Comparative Evaluation of Mobile Platforms for Non-Structured Environments and Performance Requirements Identification for Forest Clearing Applications

Abstract

The effort to automate is present across all industries. It has an economic purpose but potential impacts go far beyond economics. Research has been carried out and a lot of investment has been made in automation in a variety of industries, as well as in agriculture and forestry, which resulted in efficient solutions for diverse applications. In fact, more solutions have emerged in the field of agriculture than in any other. This can be explained in economic terms, but also in light of the complex navigation required because of unstructured environments such as forests. This paper provides a comprehensive review of existing mobile platforms and presents a comparative study for an application in forest clearing. We evaluate the size, automation levels, traction, energy source, locomotion systems, sensors/actuators availability and tools that such an application must have to succeed in its function. Hence, it will be possible to evaluate the feasibility of retrofitting an existing platform into an electric unmanned ground vehicle for forest clearing or if it is easier to start development from scratch. The evaluation results reveal that an electric unmanned ground vehicle for forest clearing is currently unavailable in the market and that a new platform is needed. The performance requirements for such a platform are identified and proposed in the paper.

1. Introduction

The panorama of destruction forest fires leave each year globally is well-known due to their strong impact on society and on the environment. Apart from the natural environment destruction, the worst consequences of the wildfires are the loss of human lives, wildlife and the destruction of property, including vehicles, houses and other infrastructure. High voltage lines were identified the source of ignition in forest fires, caused by strong winds that make the power lines touch each other or touch the surrounding vegetation [1]. Having the ground under the power lines clear is important to prevent a small fire from spreading. Regardless of the ignition source, for a wildfire to spread, fuel is needed, which together with oxygen completes the well-known fire triangle [2]. Therefore, one of the best measures to prevent the beginning and rapid spread of wildfires is the absence of combustibles. Clearing the vegetation diminishes the power and ability of fire to progress in an increasingly adverse climate [3]. Electric lines are structures that extend over great distances, crossing all kinds of terrains. Clearing the strip of land directly below these is a huge challenge which must to be undertaken once or twice a year. Portugal alone has around 9000 km of high voltage aerial electrical lines [4]. The cleaning and maintenance of the land under the power lines is mandatory and the responsibility of the company that operates the power grid. In Portugal, the government published a law (Decreto-Lei no. 124/2006) to minimise the impact of wildfires [5]. The law stipulates, among other complex management rules, the mandatory clearance of forest vegetation by landowners with strict deadlines and fines for failing to do so after the great fires of 2017 [6]. However, clearing is an expensive and slow task, often performed with inefficient and insufficient methods. It is still very common to clear the forest manually using chainsaws and brush cutters. In larger areas this requires a vast human workforce, as it takes a lot of time and exposes workers to dangerous conditions [7].

Clearly by machines is possible, it has limitations. It is necessary to guarantee access to the operating sites, safety conditions and specialized personnel to operate the equipment. Small landowners often cannot afford the expense. In large commercially exploited forest areas, on the other hand, clearing is performed with large forest mobile industrial machines (MIMs) [8]. These MIMs can be environmentally very aggressive, disturbing the fauna and flora with noise, producing vibration, compacting and spreading dust and several pollutants in the air and soil [9]. For these reasons, it is necessary to develop solutions that are economically viable for small forest owners and organizations, environmentally conscious, versatile and easy to use.

Electric vehicles have received much attention and have seen a drastic increase in importance during the last 10 years [10]. Although electric vehicles (cars, buses, two- and three-wheel vehicles) have already proved their value for urban and suburban transportation, in rural areas, where activities require high traction force and long working hours, the main power source is still combustion engines using fossil fuels [11]. In activities such as agriculture, forestry and construction, electric MIMs are being progressively introduced with relative success [12], but there is still no unmanned ground vehicle for forest clearing (UGV4FC) available.

In this work, the feasibility of two scenarios is studied: first, retrofitting a MIM into a UGV4FC and second, the development of a UGV4FC from a collection of existing systems. The term “retrofitting” refers to adding new engineering features to an older model machine. In this case, retrofitting will enable the use of older generation MIMs, preventing condemning all to obsolescence until more modern models become available [13]. This solution is very interesting from the point of view of sustainability and reuse of resources and therefore must be studied. On the other hand, the development of new UGV4Fs is inevitable; therefore, it makes sense to study proposed methods. Developing a UGV4FC is a very complex task. A UGV4FC is made up of several systems which are made up of hundreds if not thousands of components. Each system needs to be individually designed, manufactured and tested. The hydraulics, the mechanics and the electronics are designed and manufactured by different companies spread all over the world that need to work together in harmony. There are dependencies that have to be resolved, logistics that have to be set up and processes that have to be defined, tested and put into practice. All this takes years to develop and integrate and involves hundreds of people and many resources. What we propose in this work is to develop a UGV4FC in a different way, which involves jumping over the entire solution design and conceptualisation phase and focusing only on what has not yet been created using systems that are already made, tested and proven. The study begins with the definition of the use scenario followed by the definition of the key performance indicator (KPI) characteristics that will make the developed UGV4FC efficient and competitive. The scenario corresponds to the forest environment where the MIM will operate. The set of specifications was achieved by comparing several existing MIMs and analyzing the suitability of its different systems to the specific characteristics of the forest clearing operation. This methodology allowed us to establish the best specifications for the development of the UGV4FC. The detailed comparative study of existing mobile industrial machines for forest clearing in terms of their size, automation levels, traction, energy source, locomotion systems, sensors/actuators availability and tools, resulted in an assessment of what characteristics they must have to succeed in their function. This approach not only makes it possible to compare different MIMs from different areas, but also allows the continuous addition of new MIMs to refine the ideal characteristics of the machine that is intended to be developed.

This assessment is available at https://ipleiria-robotics.github.io/MIM_AC_UGV4FC accessed on 15 September 2022 [14].

1.1. Usage Scenario

The base scenario for the UGV4FC operation is a firebreak that may or may not be directly under a high voltage power line. These areas can present a high slope, rocky soils and bushes that can reach 1 m in height and on average 15 mm in diameter prior to cleaning. These land strips can be up to 50 m wide and are flanked by trees or vegetation with heights that vary between 1 and 20 m. The strips are intercepted several times by roads and forest paths. Figure 1 shows a typical firebreak strip in central Portugal.

1.2. Selected Mobile Industrial Machines

Table 1 in Section 4 contains the identified relevant forest MIMs chosen to serve as the technological baseline for this study. These MIMs are very diverse, meaning that they have different propulsion systems, locomotion, steering and control modes in addition to different levels of automation. Most of the selected MIMs are for agricultural applications, in which more solutions are available but some construction and forestry MIMs were also included.

2. State of the Art

2.1. Forestry

In forestry, the evolution towards the use of heavy machines started around the early 1980s. By the late 1990s, 95% of the wood harvested in the Nordic European countries was already cut with machinery [15].

Over the years, a huge effort and investment has been made to increase efficiency, productivity, safety and create better working conditions for operators through the development of increasingly automated equipment [16].

2.1.1. Timber Industry

The timber industry operates in man-made organized forests. Here cut-to-length systems dominate, with a single-grip harvester, debranchers and crosscutters turning trees into logs. A special-purpose vehicle, designated as forwarder, is then used to pick up the processed logs from the ground and transport them to a landing area close to a road accessible by timber-trucks. These is a very efficient and productive industry thanks to these technically advanced machines [17]. In terms of autonomy, the most advanced equipment is not fully autonomous, needing assistance from a human operator. Fully autonomous systems are currently seen as highly unlikely in the short-term due to the many complex decisions that have to be made, such as which tree to cut, in which direction, where to position each machine and where to place the logs [18]. Unmanned harvesters remotely controlled by the forwarder human operator, loading directly onto the bunk is where the evolution to a fully automated system stands today [19].

2.1.2. Fuel Management in Forest Areas

There is a proven relationship between the climate changes that are taking place and the increase in the number and severity of wild fires across the globe [20]. Wildfires are taking place in geographies that were not previously affected [21], occur more often with larger dimension and consequently are more difficult to fight. In addition to the hotter and drier climate, the existence of vegetation on the forest floor makes it even more difficult to stop fire progression once it gains some dimension. Small trees, fast-growing invasive species, bushes and other combustibls materials that accumulate on the ground are the fuse and fuel of major fires, hence the importance of their management [22].

Fuel management is particularly difficult in wild or poorly organised forests. This is the typical scenario of forests that have no industrial purpose [23]. It ends up being in these forests that the biggest fires occur with the greatest material losses and sometimes human losses as well.

As previously mentioned, the cleaning of non-industrial forests is essentially performed manually. The MIMs used are mostly tractors adapted from agriculture. Just very recently, small remotely controlled machines equipped with mulchers began to be commercialised [24]. These are the most advanced automation technology available on the market until now.

2.2. Forest Management of the Future

2.2.1. Forest Automation Challenges

Aiming at fully autonomous systems, over the past two decades, researchers have been developing systems and mechanisms in which the human operator and the machine work together. Such research has addressed the levels of automation available for handling the various aspects and challenges that operating in a forest implies such as data acquisition, information analysis, decision making and the action itself. The capacity of perception, reasoning and action of a human being is hard to match with robotic systems, but the capabilities of the human and the robot can complement each other [25]. Human–robot systems have been the subject of constant research, and the results have taken many shapes and forms, evolving from continuous human-controlled master–slave mechanisms to robots incorporating artificial intelligence under supervision of a human operator [26]. Intervention of a human in the operation loop generally improves the performance of the global system by increasing guidance accuracy, enhancing target identification, shortening processing time, reducing system complexity and handling unknown and unpredictable events that fully autonomous systems cannot deal with [27].

2.2.2. Robot Swarms

One day forests will be managed by swarms of organized robots responsible for keeping the forest clean and healthy. Robot swarms are groups of robots that act autonomously based on only local perception and coordination with neighboring robots. The collective behavior emerging from the self-organized interactions between the many robots of a swarm will allow them to solve complex tasks [28]. The main benefits of this approach are scalability and reliability. Robot swarms will continuously monitor and maintain the forest health. They can be used as watchdogs and to fight diseases and pests, as well as inhibitors of irresponsible behavior such as dumping garbage in forest areas and misuse of fire.

3. Key Metrics for the Development of a Forest Clearing Machine

In this section, the key features that a Mobile Industrial Machine (MIM) must have to perform the forest clearing operation are defined and analyzed. To assess which characteristics best meet the requirements imposed by the forest clearing function in the previously established scenario, the selected equipment will be compared in terms of its mobility, energy source, type of control, perception, communications, tools and availability. For this purpose, each compared parameter received its own adequacy scale that varies between 1 and 5, with 1 being not adequate and 5 being very adequate.

3.1. Mobility

The Society of Automotive Engineers defined vehicle mobility as “the ability of vehicle to transverse a terrain”. In ground robotics, vehicle mobility is better defined by the ability of the robot to interact with the ground maintaining a safe trajectory and defined orientation at a certain speed [29]. The various subsystems that contribute to mobility will be analyzed below, and their suitability for the scenario created above assessed.

3.1.1. Locomotion

The locomotion subsystem of an mobile platform (MP) plays a key role in achieving mobility. Robotic vehicles have been developed in different shapes, sizes and configurations [30]. The main types of locomotion systems are wheeled, tracked, hybrid (a mixture of the first two) and legged (Figure 2). The locomotion system is defined considering the application or function within the application that the system will execute. The characteristics of the terrain, its orography, the type of soil, the number, shape and type of obstacles, are just some of the aspects to be considered when defining the locomotion system [31]. According to the established scenario, the locomotion system must be able to overcome loose terrain, both dry and humid, covered with vegetation, rocks, branches, slopes and other obstacles of varying dimensions. It must be delicate and at the same time strong and durable, cost effective and easy to maintain. A legged system of locomotion can be very precise, cause little ground disturbance and literally allow walking through complex terrain. The independent control of the legs (supports) makes this type of locomotion very stable even in unconventional poses. The other systems are simpler but also less effective in rough terrain, especially if it is inclined and has loose elements such as rocks and branches [32]. The use of tracks instead of wheels allows increasing the traction capacity and distribute the forces minimising the pressure exerted on the ground. These systems are simple to manufacture, maintain and control [33]. The track can have different shapes and be made of different materials, which allows modelling the traction and the effect of the weight over the ground and surface organisms such as roots. Wheeled platforms are used in the forest but normally do not leave the roads, as they quickly lose traction if any wheel or axle is not in contact with the ground. They are therefore more used to transport material out of the forest or in machines with very complex suspension and steering systems. The suitability scale according to the established scenario and the challenges it presents in terms of locomotion strategy is as follows:

	Wheels	Tracks	Hybrid	Legs
Evaluation	2	3	4	5

3.1.2. Steering Type

There are also several steering systems for anMP. The most common are conventional, articulated, coordinated steer, skid steer and independent (Figure 3). The steering system type largely depends on the application and may be determined or conditioned by the locomotion system [34]. This happens in vehicles with two fixed tracks that have a skid steer type steering, also known as differential steering such as the MIM in Figure 3c.

Independent steering allows all kinds of movements in any direction at any time in highly complex terrain. This is the type of steering used in rovers such as Nasa Mars Curiosity rover [35]. They are not very common on MIMs due to the difficulty to control and the need for a complex suspension system to keep all wheels on the ground. Articulated steering systems are very common in large forestry machines [36]. Machines whose locomotion system is based on wheels, tracks or even both, use this type of steering to improve agility and traction in rough terrain. Skid steer or differential steering is used on rigid-wheeled or tracked-based systems [37]. Its agility and control simplicity make this steering system very common in small and large agricultural, construction and forestry machines. Coordinated and conventional steering systems are mostly used in wheel-based locomotion systems. These are not particularly agile, but they are quite common. Transport trucks and log handling machines are just two examples of its use in forestry applications. The suitability scale according to the established scenario and the challenges it presents in terms of steering method are as follows:

	Conventional	Coordinated   steer	Skid steer	Articulated	Independent
Evaluation	1	2	3	4	5

Figure 3. Key metrics—steering: (a) conventional; (b) coordinated steer; (c) skid steer; (d) articulated; (e) independent.

Figure 3. Key metrics—steering: (a) conventional; (b) coordinated steer; (c) skid steer; (d) articulated; (e) independent.

3.1.3. Gross Weight

Weight is critical to the mobility of any MP. Too much weight impairs mobility just as too little weight can limit functionality. Lighter machines do not support tools capable of clearing the vegetation, while heavier machines are less agile to navigate, harder to transport and generally more dangerous. The effect of the MP weight on the ground also must be considered. According to the literature [38], to ensure ecological compatibility the maximum ground pressure allowed in the soil varies between 0.025 MPa on weak forest soils and 0.07 MPa on moderate forest soils. A tracked MP exerts less pressure on the ground than anMP with wheels of the same weight [39]. Considering the functionality which is discussed further and the effect of the weight on the forest soil, the MP weight must not exceed 2500 kg. Depending on the width of the track and its material, the MP will exert pressures perfectly in line with the maximum values allowed, preserving not only the soil under the surface as well as the soil of the surface where some tree roots pass.

3.1.4. Payload

The payload corresponds to the weight that the MP can carry in addition to its own weight (see Figure 4a). This load adds to the overall weight with the implications for mobility that have already been noted. A large payload capacity allows the use of more systems (e.g., more sensors) and larger more powerful tools (for faster clearing). For forest clearing, the most common tool used is the mulcher. Its size and weight determine the mulching capacity and the time it takes to clean a certain area [40]. The greater the payload capacity of the MP is, the better it functions.

3.1.5. Turning Radius

The turning radius corresponds to the radius of the arc described by the center of the path made by the outside front wheel (or outside track) of an MP when making its shortest complete turn [41]. In other words, it corresponds to the capacity of an MP to reverse its direction of travel in the smallest possible space (see Figure 4b). This ability is very important in tight spaces such as in forests and construction sites. According to the established scenario and the challenges it presents, the turning radius is important as it allows getting around obstacles and navigating through intricate spaces, such as the bases of the electrical power lines towers.

3.1.6. Ground Clearance

In a simplified way corresponds to the distance between the underside of the MP and the ground. It is a crucial characteristic for mobility because it partly defines the MP ability to overcome [42]. According to the established scenario and the challenges it presents the ground clearance must be greater than 250 mm height from which most obstacles such as stones are avoided (see Figure 5a).

3.1.7. Max Lateral Slope Angle

The maximum lateral slope angle corresponds to the maximum inclination that an MP can traverse on a course perpendicular to the direction of the slope without overturning [43]. Factors such as weight, width and position of the centre of gravity are critical to travel along a hill (see Figure 5a). This type of movement is very common in forestry applications.

3.1.8. Max Climbing Angle

The maximum climbing angle is the angle from which the MP cannot progress on sloping terrain due to loss of traction [44]. Factors such as weight and the locomotion system are decisive in the maximum climbing angle that a vehicle can overcome (see Figure 5b). In the forest, the ability to climb high slopes is essential as many of the forested areas are located on hillsides.

3.1.9. Angles of Approach and Departure

The approach and departure angles are the maximum angles that an MP can overcome when approaching and exiting an inclined terrain (see Figure 5c). In the forest, the change in slope can be very sudden; therefore, high approach and departure angles are extremely important to travel in these places [45].

3.2. Power

Diesel is the most common source of energy in agricultural, construction and forestry MIMs. However, with the introduction of increasingly automated vehicles, the number of electric MIMs has been growing [46].

3.2.1. Propulsion System

The type of propulsion is related with the time of use and the power required for the application. It is also necessary to consider the use and consequent power consumption of auxiliary systems, such as control systems, tools and attachments. MIMs have been equipped with high-power diesel engines known for their robustness and reliability when subjected to heavy duty tasks in harsh environments [47]. More recently, work has been undertaken towards hybrid electric systems as alternative propulsion units. These allow increasing the overall efficiency of the MIMs, reducing the amount of pollutants produced per unit of work. Full electrification of MIMs is increasingly seen as the way to further reduce emissions, reduce dependence on fossil fuels and boost automation [48]. For the development of a UGV4FC, it makes sense to focus on power solutions that are environmentally friendly and that at the same time facilitate the automation of processes. Suitability scale according to the established scenario and the challenges it presents in terms of propulsion system follow:

	Petroleum-derived   fuels	Hybrids	Full-electric
Evaluation	1	3	5

3.2.2. Operational Range

The operational range is a very important parameter as it determines the time that a certain MP works with one tank of fuel or battery charge if it is electric. For certain applications, a large operating range is necessary due to the effort or distance to be covered. Other applications are not so demanding and even allow breaks for re-charging or refueling [49]. Operation in the forest presents several logistical challenges, with a large operational range meaning more hours of uninterrupted work which allows a lower operational cost and greater efficiency. The UGV4FC ideal operational range would be around 8 h without interruptions. To achieve this performance, it is important to select the proper battery technology and/or define the most suitable charging strategy for the application [50].

3.2.3. Battery Technology

Advances in battery technology have been vital to the development of a wide range of electric propulsion systems, such as autonomous vehicles, robots and drones [51].

• Lithium-ion (Li-ion) batteries have gained substantial improvements in the last decades, first to support portable electronics and more recently to enable cost-effective high performance electric vehicles [52]. The rapid decline in costs is mainly due to a massive increase in production scale and the increase in cell performance, making cells cheaper on a cost/kWh basis. Lithium-ion batteries have high energy density (reaching 250 Wh/kg in 2022), offer a large number of cycles and good thermal and chemical stability, leading to lower life cycle cost when compared to other battery technologies, reason why they are widely applied in electric vehicles.
• Nickel-metal hydride (Ni-MH) batteries, in comparison with lead-acid batteries, have up to double the specific energy and a greater energy density [53]. However, Ni-MH batteries do have disadvantages, such as having lower charging efficiencies than other batteries and a higher self-discharge that is aggravated in a high-temperature environment.
• Solid-state batteries (SSBs) are considered the next-generation energy storage technology [54]. Unlike lithium-ion batteries, SSBs do not need extensive monitoring, cell balancing or cooling systems. They have more energy density (up two times more than lithium-ion batteries), and they are smaller, simpler and lighter energy storage systems. They promise higher energy density, wider operating temperature range and improved safety for electric vehicles. Lithium metal anodes (Li), with sulfide-based solid-state electrolytes (SSE) and nickel-rich cathodes are the most promising compositions. However, the battery cycle life at high cathode mass loading and high current is still limited because the failure mechanism is not fully understood. This is preventing the technology to scale up to the sizes required for electric vehicles [55].
• Lead-acid batteries were the first rechargeable batteries ever made. Although the technology is outdated, it has stood the test of time and is still among today’s most widely used types. It is popular due to its low cost and ability to operate efficiently even at low temperatures, which often prevails over their low energy densities and low life cycle times [56]. Lead-acid batteries also present potential environmental problems if they are not properly disposed of or recycled.

3.2.4. Charging Strategy

There are two main strategies for charging an electric vehicle’s batteries: “opportunity charging" or “battery swapping". In opportunity charging, the batteries are charged several times during the working hours, while in battery swapping the battery is used until it is necessary to substitute it with a fully charged one. In forestry applications, both opportunity charging and battery swapping may be feasible. Opportunity charging became an option if the MIM operating range allows charging to take place outside working hours, or if the fast-charging option exists and is so fast that it makes sense to interrupt the operation to recharge [57]. Battery swapping, on the other hand, makes sense if the swap operation is simple and fast. Either way, these two forms of charging the batteries are particularly suitable if the electricity that charges the batteries comes from a clean or renewable source.

3.2.5. Charging Time and Swap Time

The time it takes to charge a battery can be a determining factor when choosing a machine for a demanding operation such as forest clearing [58]. Different battery technologies have different charging times, which is a determining factor in choosing the ideal technology for a given application. The charging times of an electric vehicle do not compare with the time of filling the fuel tank in an internal combustion vehicle, at least with the present-day battery technologies. Swapping batteries also presents some challenges. Batteries are usually heavy and therefore difficult to maneuver. The swap must be made by some type of mechanism that is not exactly at the place of the cleaning operation, which implies some travel distance to change the battery (see Figure 6). In the forest, the Unmanned Ground Vehicle for Forest Clearing (UGV4FC) will sometimes be far from the charging point and/or the battery swap point, but once there, swapping batteries can have advantages as this operation takes much less time to perform.

3.2.6. Available Power

The available mechanical power is the sum of all the power sources of an MIM. In a conventional internal combustion engine MIMs, the available power is produced by the internal combustion engine, which then feeds the remaining systems. In electric MIMs, the power distribution can be centralized or decentralized. In a centralized system, the available power is the power produced by a single electric motor, similar to what happens with a conventional internal combustion engine MIM. In a decentralized power distribution, each system has its own electrical motor and actuators. So electrification is not just about using batteries as the power source. It is also about using electrical drives to replace engines and hydraulics. Electric motors have huge torque at low speeds, they are more efficient, more reliable and lighter; however, all systems will use the same limited energy source, which must be taken into account when sizing the battery system [59].

3.2.7. Hydraulic Power

The hydraulic system can be independent or connected to the MIM main power system. Either way, there must be enough power for it to work whenever it is necessary. The size and capacity of the hydraulic pump are the characteristics that define the power of the hydraulic system. The request can be permanent, if all actuators are hydraulic, or intermittent. The size of the system and the demands on the actuators define the power required for the hydraulic system to work [60]. The attachment used in a UGV4FC is very demanding in terms of torque, so it would make sense to be hydraulic and therefore have a hydraulic system in the UGV4FC. Its power must be such that it can handle the most powerful attachment possible to fit the UGV4FC.

3.3. Control

Degree of Autonomy

• Manual: This includes the range of machines and tools that depend on the human operator at all levels, from position and attitude control to decision making. The equipment has no ability to give feedback to the operator or adapt to changing conditions. The human operator is at the heart of the operation and therefore at greatest risk while operating the equipment [61].
• Remote-controlled: This involves remote operation by means of a remote control as shown in Figure 7a. It allows the operator to be taken away from the action, which gives him greater protection against any type of dangerous event in the vicinity of the operation [62]. These MPs may have sensors and safety systems that alert the operator to events that may be unknown to him and that may lead to some dangerous operation.
• Operator-assisted autonomy: Operator-assisted autonomy allows operators to focus on the operation as their MIM drives itself. While MIMs are still manned by an operator, guidance technologies help maintain row-to-row MIM accuracy in the field in order to reduce overlaps and skips. This category of automation uses both vehicle and environmental data to develop an information hub [63]. An MIM equipped with ISOBUS Class 3 functionality falls into this category as this technology allows, among others, the control of tractor ground speed and power take-off (PTO) functions. Satellite imagery and soil sampling maps help in the coordination and optimization of in-field path plans, which fall in the category of operator-assisted autonomy. These technologies help save energy, decrease labor costs and reduce operator fatigue while also maximizing efficiency.
• Supervised autonomy: The supervised autonomy category of automation allows providing in-field supervision while the unmanned MIM performs designated tasks [64]. With an unmanned MIM, supervised autonomy takes productivity and efficiency to a new level by enabling would-be-operators to monitor the performance of their MIM while simultaneously accomplishing other strategic, in-field tasks. MIMs with this autonomous capability are equipped with high-precision GNSS and intermediate level sensing and perception to avoid environmental obstacles.
• Full autonomy: Full autonomy is the most advanced category of automation [65]. Full autonomy allows an MIM to be operated with remote supervision—such as from the farm office—or via artificial intelligence control systems. Additionally, an MIM with full autonomy has the ability to account for weather and moisture levels, which serve to further increase productivity and efficiency of the farming operation.

3.4. Perception

Perception is vital for an MIM to acquire knowledge about its work environment and itself [66]. This is achieved by means of sensors and subsequently extracting relevant information from those sensors’ measurements (see Figure 7b).

3.4.1. Position and Attitude Sensors

Position sensors are devices that can detect the movement of an object or determine its relative position measured from an established reference point. Attitude sensors are instruments for determining the angular deviations of the axes of an object from preset directions. A combination of accelerometers, gyroscopes and magnetometers measure machine’s specific force, angular rate and orientation. The magnetometer is commonly used as a heading reference sensor [67]. These sensors are often combined in what is called an inertial measurement unit (IMU). Typical configurations of Inertial Measurement Unit (IMU) contain one accelerometer, gyro and magnetometer per axis for each of the three principal axes.

3.4.2. Application-Specific Sensors

Force torque sensors are used to know what force/torque is being applied at a particular point. They are especially useful in situations where force limitation is necessary. They can also be found in force feedback systems [68]. Contact and proximity sensors detect the presence of objects and measure their proximity. Cameras are used in the identification and selection of crops ready to be harvested in real time and in sorting systems. Humidity, temperature and specific gas detection sensors are also used together or separately for automatic process control and activation of specific functions on board the machines.

3.4.3. Awareness and Safety Sensor Systems

Awareness and safety systems are not composed of a particular type of sensor, as they usually involve complex algorithms as well as heterogeneous sensor fusion [69] based on a variety of sensors, which are application-specific. These systems are used to prevent dangerous situations from occurring, namely the detection of people and/or animals around the MIMs, obstacle detection, the adaptation to the ground conditions and the blocking or activation of specific systems [70].

3.5. Communications

Using different techniques, communications occur in two ways, intra-agent (internal communications) and inter-agent (external communications) (see Figure 7a). Intra-agent communications concern communications between modules on a single MIM, while inter-agent communications occur between MIMs and infrastructure [71]. Intra-machine communications can be high bandwidth, with low latency and high reliability. Inter-machine communication may suffer from reliability problems due to radio interference.

3.5.1. Internal Communications

Modern MIMs, such as tractors or highly automated harvesters, employ numerous controller area network (CAN) networks to enable communication within the control system, such as engine Controller Area Network (CAN) and vehicle CAN. Due to the growing number of subsystems that these MIMs have, a specific communication protocol called, ISOBUS based in CAN was developed. ISOBUS was introduced to standardize communications so that any tool or attachment can be added and controlled by the MIM regardless of its manufacturer [72].

3.5.2. External Communications

Connectivity technologies are an essential part of modern MIMs, which utilizes various digital technologies to improve the efficiency of operations. Emerging communication technologies—5G [73], sensor networks, satellite systems and tactical networks as wireless backbones—provide greater control, while the operator is freed from less productive functions such as driving.

3.6. Tools or Attachments

Tools on agricultural and forestry MIMs are called attachments (see Figure 8a), allowing the machine to perform specific functions. In agriculture, MIMs generally have a high diversity of attachments, while in forestry and construction, it is more common for each machine to have only one or two functions and therefore fewer attachments. The limit of the attachment number is related to the weight of the MIM and the Power Take-Off (PTO) characteristics [74]. There are International Organization for Standardization (ISO) standards that regulate the various aspects of attachments, such as their fitting and connection to the PTO, therefore, there is interchangeability of attachments between different manufacturers. The main attachment for forest clearing is the mulcher, although there are other attachments that are complementary or that can replace the mulcher in certain situations. The mulcher can be actuated by the PTO or independently by its own motor [75]. The mulcher can be attached directly to the front, to the rear, or to the end of an articulated arm for clearing in hard-to-reach places. The mulcher itself is composed of a steel cylinder with teeth called hammers that rotate against the bush, breaking it into small pieces as the machine advances. The standardized fittings, the allocated power and its size are relevant factors with regard to tools and attachments when developing an UGV4FC.

3.7. Availability and Price

The availability of equipment for clearing the forest is relatively large if we consider that this operation can be carried out by MIMs normally used in agriculture and construction. Specific solutions for clearing the forest are relatively recent and are already on the market. As the level of autonomy goes up, there are fewer solutions on the market, and their prices rise sharply [76]. MIMs for clearing the forest and for the timber industry with a high degree of autonomy are still in academia or concepts without validation (see Figure 8b).

4. Comparative Analysis of Different Platforms

After characterizing the most relevant parameters of a mobile robotic platform for use in demanding outdoor environments, such as forests, a set of MIMs was selected to be compared. The aim is to identify the most suitable MIM platform for retrofitting and in the process identify the key performance indicators of a UGV4FC. In this section, the MIMs are compared in terms of key performance indicators (KPIs), that is, their mobility, power, control, perception, communication, tools and availability. The MIM’s parameter selection criterion at this stage was based on a macro-assessment of its suitability for work in the forest environment.

4.1. Overall Results

In this section, the results in each of the feature groups are discussed, and the set of features of the machine are defined. The graphs in Figure 9, Figure 10 and Figure 11 contains the analysis performed on 34 characteristics divided into 7 groups identified as key performance indicators of a UGV4FC.

4.1.1. Mobility

Four MIM stand out in terms of mobility KPI:

All of them were developed or have versions with forestry applications in mind and therefore have the appropriate characteristics to move in this type of terrain. Three of them use tracks and a skid steer steering system, while the RakkaTec Rakka 3000 [77] uses wheels but compensates with an articulated steering (typical configuration of a forest harvester). Weight is a characteristic that greatly impairs mobility as well as the general dimension of the equipment. The LTU&TFP [78] was penalized, despite being a forest machine, because its size is not suitable for forest cleaning, but for cutting and transporting trees, always acting on the periphery of the forest and not in its interior. The Milrem Multiscope [79] stands out from the McConnel Robocut [80] and the Energreen Robomax [81] and other similar machines such as the Green Climber LV600 [82] due to its higher tracks and configurable chassis.

4.1.2. Power

In terms of power KPI seven machines distinguished themselves:

Most have electric motors. Those that are not electric have either a hybrid engine, namely the Milrem Multiscope, or an extremely powerful and efficient diesel engine, such as the Kubota X Tractor [83]. The most used battery technology is Lithium-ion, with the exception of Kovaco Elise900 [84], which uses lead-acid batteries. The Zyelektrikli [85] is also worth mentioning because it is the only 100% electric conventional agricultural tractor of a considerable size available in the group. The Monarch MK-V [86] distinguishes itself by presenting the option to swap batteries and opportunity charging.

4.1.3. Control

Six MIMs have had notable marks in terms of control KPI include:

In terms of control, equipment with a high degree of autonomy and control systems more easily adaptable to the forest environment stood out from the others. The LTU&TFP and the Volvo LX03 [87] are both technologically very advanced machines. They are also both prototypes, however in different states. The Volvo LX03 is construction equipment, while the LTU&TFP is forestry equipment. However, the Volvo LX03 has a more mature control system with several autonomous functions both in terms of the operation for which it was developed and in terms of navigation.

4.1.4. Perception

In terms of the perception KPI there are 10 prominent MIMs:

The machines that stood out are those with the largest number of sensors. Despite not being fully autonomous, the Monarch MK-V is equipped with a series of specific sensors which allow it to predict near future environment conditions. This sensing capability is also essential for forest clearing.

4.1.5. Communications

Eleven MIMs distinguished themselves in terms of communications KPI:

Agricultural machines stood out in terms of communication. Not only do they have a huge number of internal systems, but they increasingly communicate with each other in joint operations, as they communicate with the cloud to obtain and send data, whether positional or meteorological. All of these devices use the same communication protocols that were developed specifically for agriculture operational MIMs. LTU&TFP, RakkaTec Rakka 3000 and Volvo LX03 were not developed for agriculture but also have advanced communications systems. The forest environment creates some communication challenges as operations can take place in places without network coverage that are densely forested. The LTU&TFP MIM is probably the only one that has made developments in this direction since it is a research project specifically developed for the forest.

4.1.6. Tools and Attachments

Regarding tools and attachments KPI, seven MIMs are worth a mention:

As in the previous KPI (communications), the agricultural MIMs also stand out. The largest number of attachments are made to adapt to the standard connections of agricultural equipment: Cat 1 and Cat 2 and 3-Point Hitch. There are also ISO standards for the other MIMs such as ISO 24410:2020 for earth-moving machinery such as skid steer loaders, but there are not nearly the same number of solutions. Some machines for more specific use, such as the H&H Thermite RS1 [88] and the Shark-Robotics Colossus [89] only use proprietary tools dedicated to their function and therefore do not score very well in this parameter. The AgXeed AgBot [90] stands out because it has a three-point front linkage Cat 2 and a three-point rear linkage Cat 3, like an agricultural tractor, and a powerful electrical PTO.

4.1.7. Availability and Price

Five MIMs distinguished themselves in terms of availability and price KPIs:

It was precisely the forest clearing machines that stood out in this category. In addition to availability in the market, these have a reduced cost when compared to larger agricultural machines.

4.2. Discussion

A comprehensive survey was carried out on the available platforms which can provide technical solutions to be used for forest clearing operations. Although the market is in rapid transformation towards sustainable energy solutions and increased autonomous operation capabilities in non-structured forest environments, there is not yet a commercially available off-the-shelf solution for forest clearing. On the other hand, in industries such as construction, solutions are emerging such the Bobcat T7X [91] the Hicon HIDROMEK [92] and the Volvo LX03. However, there is nothing that compares to agriculture, in particular large-scale precision agriculture, where the main brands already have highly sustainable, efficient and automated solutions on the market, such as the John Deere 8R [93], Fendt 500 Vario [94], New Holland NHDrive [95] and the Yanmar YT01 [96]. The few large equipment manufacturers that still do not have solutions on the market have, however, already presented their vision for what their equipment will be in the future, with concepts such as the Kubota X, the Massey Ferguson Next [97] and the Case IH Concept [98].

4.3. Conclusions

In this study, the critical requirements for the development of a platform for forest clearing were identified. From the analysis carried out, considering the metrics defined as important for the referred operation, it is possible to remark that both strategies are viable for the development of an unmanned ground vehicle for forest clearing. As shown in Table 1, the best candidate for a retrofit is the Monarch MK-V. This MIM stood out due to its sensory capacity and highly developed algorithms for reading environmental conditions, perceiving the surrounding spaces and acting accordingly. The Monarch MK-V MIM presents good results in practically all defined metrics, needing to undergo an adaptation to the forest environment. The retrofitting would focus on two aspects, mobility and forest tools adaptation. Figure 9, Figure 10 and Figure 11 show that other MIMs could also be adapted, some with even greater forestry vocation, such as the Bergmann M201 [99], the Raptor 100 [100] or the Milrem Multiscope. This would imply electrification, which could make such a project unfeasible depending on how the power transmission was initially engineered. The development of an electric unmanned ground vehicle from a collection of existing systems makes sense and is reasonable because there is no fully electric solution available and the number of machines that can be retrofitted is limited from the beginning. The key characteristics for the UGV4FC development were established in this work and are detailed in Table 2. This study can be further improved by adding more metrics and more MIMs but can already provide a basis for the development of a high-performance forest cleaning platform.

Ongoing research and development activities, both by several manufacturers and research institutes, are likely to achieve the required performance in the near future (up to 2025), addressing a very large potential market around the world.

Table 1. Relevant MIMs chosen to serve as technological baseline, its original field of application, development stage and evaluation.

Model	Field of Application	Development Stage	Mobility	Power	Control	Perception	Communications	Tools and Attachments	Availability and Price	Overall
AgXeed AgBot	Agriculture	CA	3.09	3.00	1.00	3.50	3.00	4.55	2.00	3.25
Bergmann M201	Forestry	CA	3.45	3.00	1.00	2.25	2.00	4.25	2.50	2.64
Bobcat T7X	Construction	CA	3.09	3.44	2.00	1.75	2.00	2.50	2.50	2.47
Case IH Concept	Agriculture	IC	2.64	3.20	2.00	4.06	4.25	4.25	2.50	3.27
Energreen Robomax	Forestry	CA	3.55	2.60	2.00	1.75	2.00	4.25	2.50	2.66
Fendt 500 Vario	Agriculture	IC	2.55	2.60	2.00	4.13	4.00	4.25	2.50	3.15
Green Climber LV600	Forestry	CA	3.36	2.60	2.00	1.75	2.00	4.00	3.00	2.67
H&H Thermite RS1	Civil protection	CA	3.27	2.60	2.00	2.00	2.00	2.50	3.00	2.48
Hicon HİDROMEK	Construction	IC	3.09	3.89	2.00	2.50	3.00	2.75	3.00	2.89
Jonh Deere 8R	Agriculture	IC	2.64	3.20	2.00	4.00	4.00	4.44	3.50	3.40
Kovaco Elise900	Construction	CA	3.36	3.67	2.00	2.75	3.00	2.50	3.50	2.97
Kubota X tractor	Agriculture	IC	2.73	4.11	2.00	4.00	4.38	4.63	3.50	3.62
LTU&TFP	Forestry	RD	3.36	3.20	2.00	3.94	4.25	4.25	4.00	3.57
M Fergunson Next	Agriculture	IC	2.64	2.60	2.50	4.06	4.38	4.63	4.00	3.54
McConnel ROBOCUT	Forestry	CA	3.55	2.60	3.00	1.75	3.00	4.25	4.00	3.16
Milrem Multiscope	Multifunctions	CA	4.09	3.67	3.50	2.50	3.00	3.00	4.00	3.39
Monarch MK-V	Agriculture	CA	2.91	4.00	3.50	4.56	4.75	4.25	4.00	4.00
New Holland NHDrive	Agriculture	IC	2.36	3.20	3.50	4.00	4.38	4.50	4.00	3.71
RakkaTec Rakka 3000	Forestry	IC	3.91	2.80	4.00	2.50	3.88	4.25	4.25	3.65
Raptor 100	Forestry	CA	3.36	2.80	4.50	1.75	2.00	4.25	4.38	3.29
Shark-Robotics Colossus	Civil protection	CA	3.36	3.22	4.50	1.75	3.00	2.25	4.50	3.23
Volvo LX03	Construction	IC	3.45	3.78	4.63	4.19	4.00	2.75	4.50	3.90
Yanmar YT01	Agriculture	IC	2.91	2.80	4.88	3.00	4.13	4.50	4.63	3.83
Zyelektrikli	Agriculture	CA	2.55	3.89	5.00	2.50	3.00	4.44	4.75	3.73

CA: commercially available; IC: industry concept; RD: research and development.

Table 1. Relevant MIMs chosen to serve as technological baseline, its original field of application, development stage and evaluation.

Model	Field of Application	Development Stage	Mobility	Power	Control	Perception	Communications	Tools and Attachments	Availability and Price	Overall
AgXeed AgBot	Agriculture	CA	3.09	3.00	1.00	3.50	3.00	4.55	2.00	3.25
Bergmann M201	Forestry	CA	3.45	3.00	1.00	2.25	2.00	4.25	2.50	2.64
Bobcat T7X	Construction	CA	3.09	3.44	2.00	1.75	2.00	2.50	2.50	2.47
Case IH Concept	Agriculture	IC	2.64	3.20	2.00	4.06	4.25	4.25	2.50	3.27
Energreen Robomax	Forestry	CA	3.55	2.60	2.00	1.75	2.00	4.25	2.50	2.66
Fendt 500 Vario	Agriculture	IC	2.55	2.60	2.00	4.13	4.00	4.25	2.50	3.15
Green Climber LV600	Forestry	CA	3.36	2.60	2.00	1.75	2.00	4.00	3.00	2.67
H&H Thermite RS1	Civil protection	CA	3.27	2.60	2.00	2.00	2.00	2.50	3.00	2.48
Hicon HİDROMEK	Construction	IC	3.09	3.89	2.00	2.50	3.00	2.75	3.00	2.89
Jonh Deere 8R	Agriculture	IC	2.64	3.20	2.00	4.00	4.00	4.44	3.50	3.40
Kovaco Elise900	Construction	CA	3.36	3.67	2.00	2.75	3.00	2.50	3.50	2.97
Kubota X tractor	Agriculture	IC	2.73	4.11	2.00	4.00	4.38	4.63	3.50	3.62
LTU&TFP	Forestry	RD	3.36	3.20	2.00	3.94	4.25	4.25	4.00	3.57
M Fergunson Next	Agriculture	IC	2.64	2.60	2.50	4.06	4.38	4.63	4.00	3.54
McConnel ROBOCUT	Forestry	CA	3.55	2.60	3.00	1.75	3.00	4.25	4.00	3.16
Milrem Multiscope	Multifunctions	CA	4.09	3.67	3.50	2.50	3.00	3.00	4.00	3.39
Monarch MK-V	Agriculture	CA	2.91	4.00	3.50	4.56	4.75	4.25	4.00	4.00
New Holland NHDrive	Agriculture	IC	2.36	3.20	3.50	4.00	4.38	4.50	4.00	3.71
RakkaTec Rakka 3000	Forestry	IC	3.91	2.80	4.00	2.50	3.88	4.25	4.25	3.65
Raptor 100	Forestry	CA	3.36	2.80	4.50	1.75	2.00	4.25	4.38	3.29
Shark-Robotics Colossus	Civil protection	CA	3.36	3.22	4.50	1.75	3.00	2.25	4.50	3.23
Volvo LX03	Construction	IC	3.45	3.78	4.63	4.19	4.00	2.75	4.50	3.90
Yanmar YT01	Agriculture	IC	2.91	2.80	4.88	3.00	4.13	4.50	4.63	3.83
Zyelektrikli	Agriculture	CA	2.55	3.89	5.00	2.50	3.00	4.44	4.75	3.73

CA: commercially available; IC: industry concept; RD: research and development.

Table 2. Characteristics identified through the use of the metrics for the development of a UGV4FC.

• Tracked locomotion	• Skid steer steering mechanism
• Electric power	• Hydraulic actuated
• Weight gross between 2500 and 3500 kg	• Minimum payload of 1000 kg
• Electric plug for electric-powered attachments	• Remote control with supervised autonomy capabilities
• Operating range over 8 h	• Charging time <4 h
• Both opportunity charging and swap battery options	• Standardized communication protocols
• Standardized coupling for attachments	• Selling price below 70 k (USD)

Table 2. Characteristics identified through the use of the metrics for the development of a UGV4FC.

• Tracked locomotion	• Skid steer steering mechanism
• Electric power	• Hydraulic actuated
• Weight gross between 2500 and 3500 kg	• Minimum payload of 1000 kg
• Electric plug for electric-powered attachments	• Remote control with supervised autonomy capabilities
• Operating range over 8 h	• Charging time <4 h
• Both opportunity charging and swap battery options	• Standardized communication protocols
• Standardized coupling for attachments	• Selling price below 70 k (USD)

CapButCapacitive Button ERMEccentric Rotating Mass FSRForce Sensing Resistor HCIHuman Computer Interaction ICIntegrated circuit LRALinear Resonant Actuator MechButMechanical Button PCAPProjected Capacity PCBPrinted Circuit Board PEAPiezoElectric Actuator PWMPulse Width Modulation RCDRemote Control Devices UXUser eXperience