Comparative Study of SLAM Techniques for UAV ^†

Abstract

In the last few decades, the main problem which has attracted the attention of researchers in the field of aerial robotics is the position estimation or Simultaneously Localization and Mapping (SLAM) of aerial vehicles where the GPS system does not work. Aerial robotics are used to perform many tasks such as rescue, transportation, search, control, monitoring, and different military operations where the performance of humans is impossible because of their vast top view and reachability anywhere. There are many different techniques and algorithms which are used to overcome the localization and mapping problem. These techniques and algorithms use different sensors such as Red Green Blue and Depth (RGBD), Light Detecting and Range (LIDAR), Ultra-Wideband (UWB) techniques, and probability-based SLAM which uses two algorithms Linear Kalman Filter (LKF) and Extended Kalman filter (EKF). LKF consists of 5 phases and this algorithm is only used for linear system problems but on the other hand, EKF algorithm is also used for non-linear system. EKF is found better than LKF due to accuracy, practicality, and efficiency while dealing SLAM problem.

1. Introduction

From the last couple of years, Artificial intelligence (AI) is became the most discussed topic in the field of robotics. It raises a great revolution in the field of research and can change the new era of the world. Its prominent results motivate the researchers to do research on it. Unmanned Aerial Vehicles (UAV) are being utilized to perform different tasks such as equipment transportation, researching, rescue, agricultural and military spy operations as well. It is an important thing of human life that makes life easy by its use. UAVs can operate anywhere, even where humans don’t have access. However, the main problems of UAVs that attract the researchers’ attention are their navigation and build-up map of an unknown environment at the same time. This problem mostly called Simultaneously Localization and Mapping (SLAM) [1,2,3,4,5,6,7,8].

The applications of UAVs are increasing with the passage of time due to their usefulness abilities such as less time consuming, reachability anywhere, transportation, etc. Therefore, mostly unmanned vehicles use prior knowledge or Global Position System (GPS) to navigate themselves and operate in the outdoor environment. However, UAVs should have a unique architecture that make these capable to operate in an unknown environment without any external positioning source.

There are many approaches available as a solution to the SLAM problem which uses different sensors such as Red Green Blue and Depth (RGBD), Light Detecting and Range (LIDAR), Ultra-Wideband (UWB) techniques, etc., and algorithms. The main objective of this review paper is to briefly and comparatively study different SLAM approaches to elaborate the best one based on their pros and cons.

2. Materials and Methods:

In this section, the different approaches are discussed to solve the SLAM problem while using RGBD, LIDAR, UWB, etc., sensors.

(A)

Ultra-wideband based Algorithm Approach:

UWB sensors-based algorithms mainly consist of two steps

In the first step, get the data about position estimation of UAV by use of UWB sensors then the collected data is used to build up the 3D map of surroundings. Therefore, this step is called Range-Only SLAM (RO-SLAM) as shown in Figure 1. The second step of UWB based algorithm includes optimization of position estimation data collected from UAV and UWB sensors for localization and mapping.

Initially, if there is no data set then it initializes randomly the value of sensors’ position rather than the re-calculation of all parameters completely as shown in Figure 2, moreover, the optimizer is used to remove the uncertainty from the values [9,10]. UWB based algorithm is appropriate for low-distance localization and mapping in an indoor environment. It is capable to acquire accurate data within a few meters or centimeters [11].

(B)

Kalman Filter (KF) based Approach:

Kalman filter-based approach is sometimes called the linear Kalman filter approach because it uses mostly for linear systems. Figure 3 shows the following five phases,

• Absolute measurement of motionless robot (A.M.O.ML).
• Absolute measurement of moving robot (A.M.O.M).
• Relative measurement of motionless robot (R.M.O.ML).
• Relative measurement of moving robot (R.M.O.M).
• Relative measurement of moving robot while position of robot not detected.

It uses linear approximation that is related to covariance error matrices and positions to find out the prior condition estimation [12]. These are mostly no-deal non-linear systems but sometimes KF can be used for non-linear systems if the first-order partial derivative of calculations is available.

(C)

Extended Kalman Filter Based Approach:

EKF is used for non-linear systems. It linearized the non-liner systems to solve the problem. Similar to other methods, it has a problem that if noise is included in the measurements then the UAV robot can divert from its original route and produce inconsistency in flight.

3. Discussion

In the last few years, Artificial Intelligence has raised a great revolution in the field of unmanned robotics and become a most discussed topic. UAVs have an importance and deep impact on human life in different aspects of daily routine. Therefore, UAVs should have architecture and design which helps them to complete their tasks in challenging environment without any external positioning source. A lot of work has been carried out to solve the problems of UAVs.

This paper represented a comparative study of SLAM approaches and briefly explained as well in the previous sections. All discussed approaches are sensors and filtered based.

The pros and cons of each approach are discussed in

Table 1. Comparison of discussed SLAM Approaches based on pros and cons.

Scheme	Approaches	Pros	Cons
1	UWB Based	Deals with indoor tasks Suitable for shorter distances	Limited range of centimeters to few meters only (not suitable for long distances) Computational cost is quite higher
2	KF Based	Deals with indoor and outdoor tasks Deals with different range of distances Computational cost is less	Does not deal with nonlinear models Does not deal with complex environment
3	EKF Based	Deals with both linear and nonlinear models Computational cost is less Loop closure efficient Deals with different range of distances	Noise and inconsistency issues while dealing with a large map

and compared to each other to find out the best one among these. Form Table 1, the comparison shows that Extended Kalman Filter (EKF) based approach has the ability to deal with non-linear models, good loop closure and is suitable for long distance as well rather than Ultra Wideband and Kalman Filter based approaches which are only capable to deal with short-distance and linear models, respectively.

4. Conclusions

In this paper, we represented a comparative study of previously existing SLAM techniques. In order to find the most suitable technique among these techniques, different parameters are set to evaluate SLAM techniques. Table 1 in the discussion section shows that EKF based techniques are better as compared to KF and UWB based techniques due to their properties of dealing with nonlinear models, having good loop closure ability, and applications in outdoor and long distances.