Monitoring of Ergonomics Score Impact on Production Processes

Abstract

This study presents the integration of ergonomics assessment data into production monitoring in a global automotive context. A cloud-based solution was developed to merge HumanTech ergonomics data with production metrics from the Cycle Time Deviation (CTD) Shift Report system via the Palantir Foundry platform. During implementation, several challenges were addressed, particularly inconsistent station naming, incompatible data formats, and duplicate or missing records. These were resolved through a harmonization process that enabled the creation of a standardized dataset. The resulting integration allows ergonomics scores to be visualized alongside cycle punctuality, takt time, and MOST-based analysis, supporting the identification of correlations between ergonomic risk and production delays. The paper also outlines the implementation steps, highlights the benefits of real-time monitoring, and discusses the potential for scalable analytics across multiple manufacturing sites.

1. Introduction

Data digitization plays a key role in increasing the transparency and clarity of production processes. This process takes place in several stages: collection, transmission, processing, distribution, and subsequent use of data in decision-making and management processes [1,2]. In order for data to be effectively usable, it must be accurate, complete, up-to-date, consistent, and correct [3,4]. Developments in the field of data management have seen a transition from conventional processing of structured data to big data analysis, which has enabled new approaches and applications supported by modern Industry 4.0 technologies. Emphasis is placed on effective management and archiving of data in order to ensure their future use [5]. High-quality data management contributes to better monitoring of business processes, increasing production quality, optimizing the supply chain, and stimulating innovation [6,7]. Analysis of this data allows for optimizing costs, increasing profitability, productivity, market share, and overall sales [8,9,10]. An example is an online solution for collecting and processing production data obtained from production and assembly processes in an automotive company. This is a process of live monitoring of the production status and subsequent evaluation through reports. Three applications are used, which are delivered in one package called Cycle Time Deviation (CTD): the Workshop application for live monitoring, the Shift Report application for evaluating data older than 24 h, and the Plant Improvement Tracker application for comparing and monitoring the impact of process changes on the analysis [11].

Data is the foundation for implementing artificial intelligence (AI) and harnessing its potential for multiple business purposes. Guardia et al. [12] identify three types of AI: analytical, assistive, and generative. They emphasize that AI can streamline the production process and add value to the final product, which emphasizes the need for new skills and roles in the industry, as well as the importance of ethical oversight for preventing misinformation and bias. Jeske et al. [13] state the opportunities of digitization and artificial intelligence, which are structured from the point of view of the technology used, work organization and personnel.

In addition to processing production data, companies are increasingly focusing on creating ergonomically optimized workplaces. The quality of the work environment directly affects the health, safety, and performance of employees. For this reason, assessing the physical and mental strain of workers is a prerequisite for ensuring an ergonomic working environment and optimizing performance. In assembly, where repetitive movements, static positions, and fine motor skills dominate, various assessment methods are applied.

Traditional manual methods are based on observation, measurement, or subjective statements. Their advantage is simplicity, low cost, and immediate applicability. These include the methods Rapid Entire Body Assessment (REBA), Rapid Upper Limb Assessment (RULA), Ovako Working Posture Analysis System (OWAS), Ergonomic Assessment Worksheet (EAWS), National Institute of Occupational Safety and Health (NIOSH), and others. These methods often form the basic framework for assessing ergonomic risks in the workplace. However, their disadvantages are their time-consuming nature, subjectivity in the assessment, and limited possibility of long-term monitoring. With the development of digitalization, intelligent software and hardware solutions are emerging to identify, eliminate, and monitor occupational risks, using automated data collection and processing with AI. These approaches deliver more accurate, continuous, and individually tailored results, but are still not the standard in manufacturing companies. These include video recordings, wearable sensors, Human Digital Twin, physiological measurements, and combined AI + IoT + Big Data systems that integrate data from multiple sources (video, sensors, physiological measurements) and are processed in real time [14].

For example, Lasota [15] presented a framework for assessing physical ergonomic hazards, while Patel et al. [16] analyzed trends in monitoring technologies, including wearable devices and connected worker platforms. Bortolini et al. [17] presented a Motion Analysis System (MAS), based on motion capture technology, that enables detailed ergonomic and productive assessment of work movements. Libanore et al. [18] supplemented the research with structured interviews with exoskeleton users. Onofrejova et al. [19] compared traditional methods (e.g., RULA, EAWS, REBA, NIOSH) with the use of wearable sensors and exoskeletons in assembly in a manufacturing environment in Slovakia. They proposed a methodology for quantitative measurement of ergonomic risk for optimizing worker workload using wearable wireless multisensor systems, Captiv and Actigraph, in selected workplaces. Capodaglio et al. [20] in a pilot study examined the short-term effects of passive exoskeletons during repetitive tasks. Other effects of exoskeletons on worker health and performance are examined in more detail in [21,22,23,24]. Peruzini et al. [25] presented a design of a human-centered industrial system, including the human factor in the digital twin. They proposed a theoretical framework for an Operator 4.0 monitoring system, which is based on collecting data on workers’ performance, actions, and reactions with the ultimate goal of improving overall productivity and production organization. The data is used to assess workers’ ergonomic performance and perceived comfort and is integrated with data obtained from machines. A combined system integrating IoT, AI, and big data analysis technology was used by Shu et al. [26] in their design of a remote monitoring and control system for oil and gas stations.

Several authors have confirmed that continuous improvement of manufacturing processes, linking the principles of Lean Manufacturing (LM) and Ergonomics and Human Factors (E&HF), contributes to increased productivity, improved working conditions [27,28,29], and reduced absenteeism [30]. Colim, A. et al. [31] analyzed the industrial implementation of a collaborative robotic workstation for assembly tasks, which was designed based on the synergistic integration of the E&HF and LM principles. Thanks to a multi-method approach (they monitored 40 positions; they used the following methods: rapid upper limb assessment, revised tension index, key indicator method, and questionnaire for monitoring worker well-being), they found that the hybrid workstation brought several benefits: reduced production times, improved ergonomic conditions, and increased worker well-being. The present paper presents the integration of two separate online applications. The first—the Cycle Time Deviation application—is focused on the collection and processing of production data. The second—HumanTech—is dedicated to the assessment of ergonomic conditions in workplaces. The aim of linking them is to identify and mitigate risks to employee health, while also analyzing the extent to which ergonomic inefficiency can affect the quality of outputs and work productivity.

2. Materials and Methods

The methodological framework of this research consists of two main stages. The first stage is dedicated to analyzing the current state—including existing procedures, applications, and the overall functioning of the process—along with a description of commonly used tools. Based on this analysis and the identification of key issues, the second stage focuses on proposing a solution. This solution aims to help users understand the impact of station ergonomics scores on cycle time punctuality, by comparing actual assembly times with projected ones.

The study is based on real-world data provided by a major global automotive company specializing in car seat manufacturing.

2.1. Current State

The current state consists of two applications. The first application, which collects ergonomics data, is from VelocityEHS–Industrial Ergonomics and is referred to as HumanTech (Ann Arbor, MI, USA). The second application, called Cycle Time Deviation (CTD), was developed internally using the Palantir Foundry (version 6.440.22) platform.

HumanTech solution: In general, the HumanTech platform is an online, web-based interface that allows users to create and manage ergonomics studies on production stations or machines. Each study can be created in one of two modes. The first mode, called Manual Whole-Body Assessment (Figure 1), is completed based on manual input from an Environmental Health and Safety (EHS) specialist and requires precise evaluation with a focus on all activities.

The second option for creating an ergonomics assessment, known as the Advanced Whole-Body Assessment, is automated and based on artificial intelligence (AI) analysis. All operator actions can be recorded, and the video is then uploaded to the HumanTech platform. After processing, the pre-built AI analyzes the footage and automatically categorizes all movements (Figure 2).

All assessments can be edited at any time, including the addition of force data, vibration information, temperature conditions, impact stress factors, or glove-related issues. One of the key advantages of the AI-assisted Advanced Whole-Body Assessment over the Manual Whole-Body Assessment is the significant reduction in data-entry errors. Whereas manual input relies on an EHS specialist to record body postures and force parameters step by step, the AI pipeline automatically extracts joint angles, segment durations, and ergonomic risk factors from video frames with consistent precision. This automation minimizes human bias and transcription mistakes, ensuring higher repeatability and reliability of ergonomics scores across multiple stations and shifts. The result of each assessment is an ergonomics station score, which reflects the difficulty of the process and its impact on human health.

The score is calculated based on key ergonomic risk factors, including posture angles, force exertion, movement frequency, task duration, and environmental conditions such as vibration or temperature. These inputs are analyzed either manually by EHS specialists or automatically through AI-based motion capture, resulting in a standardized risk value for each station.

Assessments can be updated based on process improvements and are categorized into three groups—Baseline, Projected, and Follow-up—making all modifications easily trackable.

Conducting such assessments requires solid knowledge of EHS techniques and processes; therefore, all assessments should be managed by an EHS specialist.

HumanTech employs proprietary risk assessment methods, such as Baseline Risk Identification of Ergonomic Factors (BRIEF) and Integrated Design and Ergonomic Assessment (IDEA), which quantify ergonomic risk using internally developed scoring models tailored for manufacturing operations. Unlike widely used public-domain tools like Rapid Entire Body Assessment (REBA), Rapid Upper Limb Assessment (RULA), or Ovako Working Posture Analysis System (OWAS), which rely on human-observed posture-based scoring systems, HumanTech’s methodology is built into the platform (e.g., through guided postural assessment workflows and AI-driven calculations) and cannot be directly mapped to those standard metrics. Consequently, conventional benchmark comparisons with REBA/RULA/OWAS were not applicable in this implementation.

CTD application: The Cycle Time Deviation application was developed as a production monitoring tool that tracks and analyzes various types of production data—primarily cycle times and cycle punctuality, scanner Right First Time (RFT), fastener RFT, and scrap rates. All analyses are based on different time frames. The Workshop application (Figure 3) focuses on live data that is no older than 24 h, while the Shift Report Contour analysis processes historical data up to six months old [11]. This time range is determined by system constraints: the CTD application stores a maximum of six months of historical data due to the large volume of inputs generated across more than 300 manufacturing sites worldwide, making long-term storage impractical within the current corporate infrastructure.

Currently, to assess the impact of the station ergonomics score on cycle punctuality, it is necessary to manually compare data from two different systems. The request is to develop an automated solution that connects and analyzes all data within a single application, with the possibility to extend its functionality in the future to include additional indicators if needed.

2.2. Solution Proposal

The first step was to clearly define the development scope. Based on an analysis of customer needs, we designed the solution. In the proposed approach, all data processing and analysis will be fully automated using a cloud-based system, eliminating the need for manual comparisons.

Instead, a unified standard will be implemented and applied across all manufacturing plants, ensuring greater consistency and standardization.

The goal of the development is to analyze data displayed in the Shift Report application from the CTD project, where the ergonomics score will be overlaid onto cycle punctuality, over tact, and over most statistics.

Overlaying the ergonomics station score data onto standard CTD charts allows shop floor teams to quickly visualize whether cycle time punctuality performance at the station level could be related to the associated ergonomics station score—a quantified measure of physical burden. A higher ergonomics score may indicate difficulties in completing assembly tasks within the defined or targeted time.

Feedback from many continuous improvement teams supports this feature in the shift report. In many cases, teams have manually analyzed potential trends or causes and, in numerous projects, improved cycle time punctuality by addressing station ergonomics conditions.

Before the data are analyzed, it is necessary to establish a connection between the Palantir Foundry CTD project and the HumanTech database, as both operate on separate platforms. Since HumanTech uses a closed environment without direct access to its data, a third-party cloud solution is required to stream data to the Foundry platform. For this purpose, the SNOWFLAKE solution was chosen to establish all data connections (Figure 4).

2.2.1. Data Sources

Data sources for both platforms can be divided into two groups: online data and manual data. The first group consists of automatically collected data from processes, referred to as online data, while the second group includes manually documented data, referred to as manual data. All data may be stored in multiple databases using various formats. These spreadsheet-like databases in their raw form are considered data sources.

Online data from the CTD application includes data directly gathered from Programmable Logic Controllers (PLCs) or other devices such as Automatic Inspection Devices (AIDs), fasteners, scanners, and other equipment used in assembly processes, which upload collected data into the corporate Manufacturing Execution System (MES).

Manual data from the CTD application can include time studies, shift calendars, and similar records. Manual data from HumanTech consists of studies manually entered by EHS specialists, where all body movements are recorded step by step. Manual data from the CTD application typically originates from industrial engineering activities, such as time studies or shift planning, and is often collected and maintained by local production teams. In contrast, manual data in HumanTech is structured within ergonomics protocols and entered by certified EHS specialists following specific corporate assessment procedures, which ensures consistency and expert-level authority over data quality. A more advanced approach involves uploading videos of whole-body operations, which are automatically evaluated by AI and segmented into spreadsheet cells with generated assessment results.

Since we are dealing with two different platforms, both use different data formats and storage logics. The CTD application and Shift Report, built using the Palantir Foundry solution, were developed to meet specific corporate needs based on data availability from the corporate MES, which collects data from multiple internal sources in specific formats. To evaluate data from these two systems—CTD Shift Report and HumanTech—we need to define proper data transformation methods and build customizable logic that can also accommodate future requirements.

2.2.2. Data Specification

During the phase of defining the development scope, a crucial aspect involves clearly specifying and defining the data requirements. This step is essential because the entire solution will depend on the accuracy and availability of these data. It is necessary to map out data lineages and design all computations in accordance with the formats and structures of the data that are accessible.

The CTD Shift Report application was originally developed to address specific corporate requirements, meaning that the data it handles are already organized in a standardized format preferred by the company. Building on this foundation, we have identified and selected multiple relevant datasets from the available data sources to be utilized in the solution (see

Table 1. Definition of datasets used for application development.

Datasets from CTD	Datasets from HumanTech
‘plant_mapping’	‘“PUBLIC”.“JOB_INFO_VIEW”’
‘station_mapping’	‘“PUBLIC”.“ASSESSMENT_VIEW”’
	‘“PUBLIC”.“ANALYSIS_TOOL_DATA_VIEW”’
	‘“PUBLIC”.“OU_POSITION_VIEW”’
	‘“PUBLIC”.“OU_VIEW”’

). These datasets form the backbone for subsequent analysis and reporting, ensuring that the system functions reliably and consistently across different manufacturing contexts.

By thoroughly defining and specifying the data at this early stage, we minimize potential integration issues and enable smoother data transformation and processing workflows throughout the project lifecycle.

Once the relevant datasets are successfully joined, transformed, and extracted, it will be possible to comprehensively describe the ergonomics assessment scores at multiple levels, including ‘Location’, ‘Station’, and ‘Advanced Tool’ (see Figure 5). These scores will provide detailed insights into the ergonomic conditions across different production areas and tools, enabling a more precise evaluation of their impact on worker health and operational efficiency. This structured approach ensures that the analysis is grounded in integrated and harmonized data, supporting informed decision-making and targeted improvements within the manufacturing process.

For data sorting, the ‘Location’ field will serve as the primary criterion. This field is organized into six hierarchical levels, providing a structured framework for categorizing and analyzing data efficiently across different manufacturing sites and operational units (see

Table 2. Level definitions used for data sorting in HumanTech.

Level	Description	Values
1	Fixed value used in whole corporation	Global
2	World region	Europe/Africa, North America, South America, Asia
3	Group	Just In Time (JIT), Leather, etc.
4	Site location	Kolin, Prague, etc.
5	Department	Production, Quality, Maintenance, etc.
6	Line/program	Name of the station

). By using this multi-level classification, we can ensure consistency and clarity in data management, enabling detailed tracking of ergonomics scores and other key metrics at various organizational layers. These six levels reflect the standardized organizational breakdown used across all manufacturing sites and ensure that each ergonomic assessment is accurately linked to the corresponding region, plant, department, and station in the corporate structure. This approach facilitates targeted analysis and supports better decision-making by highlighting specific areas that may require ergonomic improvements or process adjustments.

When creating new assessments, it is essential that users strictly follow the agreed-upon standard naming conventions. This consistency is vital because the naming structure plays a key role in the logic used to accurately merge and integrate data from multiple sources. Without adherence to these standards, data processing could become unreliable, leading to errors or misinterpretations in the analysis.

Implementing standardized data entry protocols was therefore a foundational step in the development of this application. By enforcing these conventions, the system ensures data integrity, facilitates seamless automation, and enables efficient data management. This approach ultimately supports accurate reporting and decision-making based on harmonized and well-structured data.

2.2.3. Data Structure

Since many of the datasets sourced from HumanTech contain essential information for application development but are not organized according to a unified standard format, it is imperative to standardize these datasets and generate the required data formats. To achieve consistency and interoperability, we will adopt the data structure and naming conventions established for the CTD application as a reference template.

To facilitate this, we will utilize the Pipeline Builder module within the Palantir Foundry platform. Pipeline Builder is a powerful tool for data integration and transformation that allows the design and automation of complex data workflows. Through this module, raw data from various HumanTech sources can be extracted and subjected to cleaning, normalization, and mapping operations to ensure the data aligns with the CTD standard format. This process includes tracking data lineage to maintain full transparency and auditability of the data sources and transformation steps.

Additionally, Pipeline Builder supports automation by enabling scheduled or event-driven workflows that process new or updated data with minimal manual intervention. The module also incorporates error handling and validation mechanisms to detect anomalies and ensure data quality before integration. By leveraging these capabilities, we ensure that the processed datasets from HumanTech become fully compatible with the CTD application’s data ecosystem, thereby supporting reliable analytics, reporting, and scalability for future needs.

2.2.4. Data Lineages

The results from ergonomics assessments conducted in HumanTech, represented as ergonomics score data, are distributed across five distinct datasets. All of these datasets require thorough cleaning and proper joining to create a unified dataset that can serve as a reliable data source for this use case. To establish and manage data lineage effectively, we utilize the Pipeline Builder within the Palantir Foundry platform.

The data lineage is built and visualized using node-based mapping, join logic tracing, and data version tracking within the Pipeline Builder tool, which allows users to audit transformations, track dataset dependencies, and verify update propagation in real time.

Within the Pipeline Builder interface, standard color codes are applied to each node to facilitate easier understanding of data lineage and to visually separate different segments of the data processing workflow. A legend explaining these color codes is located in the top right corner of the Pipeline Builder interface (see Figure 6).

To create our data lineage, we begin by importing raw datasets from SNOWFLAKE, specifically JOB_INFO_VIEW, ASSESSMENT_VIEW, and ANALYSIS_TOOL_DATA_VIEW. In the initial stages, we need to join these datasets using the join function. The selected datasets can either be copied into the Pipeline Builder workspace or accessed directly from files stored within the Foundry platform.

The first join, called ‘Assessment ID join’, merges the Assessment IDs from both raw datasets. This step is crucial for linking the actual ergonomics score with the corresponding job info ID.

In the subsequent join, referred to as ‘Join actual data’, we join the datasets using ‘Job info ID’. This operation combines all three datasets and brings together ‘Previous assessment ID’ and ‘Assessment ID’. These values may change over time; when a user modifies an assessment in the HumanTech platform, the ‘Previous assessment ID’ remains constant, while the ‘Assessment ID’ updates to the latest version. This approach allows us to track historical entries while filtering out only the most recent data (see Figure 7).

In the subsequent join operations, it is necessary to integrate the hierarchical level information sourced from the ‘OU_POSITION_VIEW’ dataset. This dataset provides detailed organizational structure data, which allows us to enrich the previously joined datasets by mapping each data entry to its corresponding organizational levels. As a result, we can assign precise numerical values representing different levels such as region, plant, department, or workstation, depending on the granularity of the data.

This process effectively “explodes” the information contained in the earlier datasets, enabling more granular analysis and reporting. By incorporating these level distinctions, we improve the ability to filter, segment, and aggregate data according to specific organizational units, thereby enhancing the overall insight into ergonomics scores and their impact across different parts of the production system (see Figure 8).

In the next join function, we transform and map each hierarchical level into a standardized, user-friendly format that can be easily understood by all standard users (see Figure 9). This step ensures consistency and clarity across the dataset, enabling more intuitive interpretation of the data.

To maintain uniformity, we strictly follow the naming conventions and level definitions outlined in Table 2. By adhering to these predefined standards, we facilitate seamless integration and comparability of data across different systems and reporting tools.

In the subsequent transformation and join functions, we merge the sorted data from HumanTech with corresponding data from the CTD application. The ‘plant_mapping’ dataset plays a key role in this process by linking the assessment location data to the precise site locations within the production network. This ensures accurate geographical alignment of the data.

A similar approach is applied in the ‘station_mapping’ join, where line and station identifiers are matched and translated into clear, descriptive names. These standardized names are essential for the frontend application, enabling users to easily identify and navigate through different production lines and stations (see Figure 10). To ensure accuracy, the mapping logic was reviewed and validated by EHS and production experts across multiple plants, using test datasets and cross-checks between station codes, location identifiers, and known ergonomic scores.

The final stage of the data lineage focuses on the last data transformations, where duplicate and null values are systematically identified and removed to ensure data quality and integrity. The resulting output is a consolidated dataset named ‘Ergo_Score,’ which is standardized according to corporate specifications.

This clean and standardized dataset serves as the foundation for overlaying station ergonomics scores within the application, enabling accurate and consistent analysis of ergonomics impact across different production stations (see Figure 11).

During the development and testing phases of the application, we identified that certain data, such as variant combinations, were initially missing. This issue was addressed and corrected during the course of the development.

With the updated logic, the application now supports filtering multiple assessments for the same station across a variety of assembly variants. This enhancement significantly expands the functionality and usability of the application, allowing users to perform more detailed and variant-specific analyses.

2.2.5. Building Schedules for Pipeline Deployment

After creating the pipeline, it is necessary to establish periodic schedules for pipeline deployment to ensure the data is regularly refreshed, as the data in both the CTD project and HumanTech platforms are continuously updated. Without these scheduled updates, the latest data will not be reflected in the frontend applications, and computations will not be performed on the most current information.

This process can be fully automated, with pipeline deployments triggered automatically based on predefined conditions. In this case, the ‘When to build’ parameter is configured to initiate a pipeline build whenever a specific resource, such as the initial raw datasets, is updated (see Figure 12). This scheduling mechanism guarantees that the data remains consistently up-to-date and accurate for all users.

2.2.6. Implementation into the CTD Shift Report Application

Once all datasets have been properly cleaned, sorted, and updated, they are ready for use in the Contour analysis. As part of this activity, it was requested to extend the existing charts in the Contour analysis module of the CTD Shift Report. Specifically, the charts targeted for extension were: ‘Cycle Time Punctuality—less punctual order’, ‘Over takt time—less punctual station order’, and ‘Over time from studies (MOST)—less punctual station order’, where CTD data are overlaid with the ergonomics score.

Since the implementation procedure is the same for all charts, this demonstration will focus on the ‘Over takt time—less punctual station order’ chart.

The first step is to open the Contour analysis application and navigate to the folder or path where the chart is located. Once the chart is found, open its settings and add an overlay layer (see Figure 13).

After adding the new overlay, it is necessary to change the dataset by selecting the ‘Ergo_score’ dataset from the data source project. Once the overlay is created and the dataset selected, the next step is to configure the overlay parameters.

First, select the chart type as ‘Vertical Bar Chart’. Set the x-axis to ‘station_name_frontend’ and the y-axis to ‘SCORE_new’ with the option ‘Exact Values’. In the options menu, set the segmenting method to ‘Grouped’. After the chart is computed and saved, the initial data visualization will be displayed.

In the formatting parameters, rename the y-axis label from ‘SCORE_new’ to ‘Ergonomics Station Score’ and adjust the color to the preferred scheme.

2.2.7. Issues Detected During Implementation

The main obstacles to standardization were related to data entry within the HumanTech platform. We identified issues with the insertion of data according to our specified levels—the six levels outlined in Table 2. Although this procedure was defined long ago, users frequently did not adhere to the standardized data entry protocols.

Another standardization challenge involved the use of obsolete assessments or reliance on manual analysis rather than employing the Advanced Tool, which is defined as the corporate standard. The use of the Advanced Tool significantly reduces human error in selecting body movements or angles and ensures that no important movements are overlooked during the assessment.

Additionally, problems with the completeness of assessments were detected. Many station assessments were either missing entirely or left incomplete, with missing data such as force measurements or other key parameters. Now that the data are consistently available on the platform, there is considerable potential for deeper investigations and enhanced analysis. Furthermore, the developed solution enables tracking of assessments on the same station, segmented by different assembly variants, providing more granular insights.

3. Results

The outcome of this research is the development of a standardized dataset that enables the integration of ergonomics station scores as an overlay on the existing CTD Shift Report charts. This dataset serves as a unified data source to enhance the analytical capabilities of the CTD platform by providing additional insight into the physical demands placed on operators during assembly tasks.

Following the successful creation and validation of the dataset, it was implemented in three key charts within the CTD Shift Report application: ‘Over time from studies (MOST)—less punctual station order’, ‘Over takt time—less punctual station order’, and ‘Cycle Time Punctuality—less punctual order’. By overlaying the ergonomics score on these charts, users can now more effectively monitor and evaluate the correlation between elevated ergonomics scores—which indicate greater physical burden—and deviations in assembly cycle times.

This integration supports continuous improvement efforts by enabling production teams and ergonomics specialists to identify stations where ergonomics challenges may be contributing to reduced punctuality or increased cycle times. Ultimately, this contributes to data-driven decision-making aimed at optimizing both worker well-being and operational efficiency.

3.1. Cycle Time Punctuality—Less Punctual Station Order

The chart features two axes: the primary axis represents cycle punctuality, which measures the adherence to scheduled assembly times based on the Maynard Operation Sequence Technique (MOST). This technique quantifies the expected time for each assembly activity. Cycle punctuality is displayed as a percentage, and the chart is ordered by stations with the lowest punctuality first, highlighting those areas with the greatest delays.

This chart is a standard visualization from the CTD Shift Report, enhanced with an overlay representing the ergonomics station score (see Figure 14). The ergonomics station score is a numerical value derived from ergonomics assessments, reflecting the physical demands and potential strain experienced by operators at each station. By integrating this score into the chart, users can more easily analyze the relationship between ergonomics challenges and cycle time performance.

Stations DAR-FI-10 and DAR-60-80 display only ergonomics assessments without any accompanying production statistics. This likely indicates that these stations are either no longer active in the assembly process or were not utilized during the evaluated time frame used to generate the production data.

3.2. Over Takt Time—Less Punctual Station Order

The chart displays four different axes. The first axis represents the over takt time summary, broken down by assembly variants. The second axis, shown as an overlay, indicates the customer takt time. The third overlay represents the expected in-process time, while the fourth overlay displays the ergonomics station score. The data is ordered once again by stations with the lowest punctuality first.

For the first three axes, the times are cumulative, accumulating cycle by cycle. The ergonomics station score is presented as a discrete numerical value (see Figure 15).

This analysis is based on the summary of over takt time cycles compared to the targeted cycle time, which is set in the corporate MES as the average time required to assemble the medium option.

3.3. Over Time from Studies (MOST)—Less Punctual Station Order

The chart displays five different axes. The first axis represents the over takt time summary, divided by assembly variants. The second axis, shown as an overlay, indicates the customer takt time. The third overlay represents the expected in-process time, while the fourth overlay shows the time studies based on the Maynard Operation Sequence Technique (MOST). The fifth axis displays the ergonomics station score. The data is ordered from the least punctual station to the most punctual.

For the first four axes, the times are cumulative, accumulating cycle by cycle. The ergonomics station score is presented as a discrete numerical value (see Figure 16).

This analysis is based on the summary of over takt time cycles compared to the MOST times—projected times required to complete any assembly and variant. From the perspective of lost time, this statistic is more precise and can also help identify areas with potential for process rebalancing.

4. Discussion

This study demonstrates a practical integration of ergonomics assessment data into a production monitoring environment through the CTD Shift Report application, with a specific focus on the challenges and benefits of implementation in a global automotive manufacturing context.

This project was carried out within the specific digital infrastructure of a global automotive company, where both the CTD and HumanTech systems are mandated corporate tools. Therefore, the purpose was not to evaluate or compare different platforms, but rather to demonstrate the integration potential and application benefits within this predefined context. The CTD application was developed internally to address production monitoring needs, and HumanTech was selected as the standard tool for ergonomic evaluation. Given these constraints, the study was structured as a practical case study rather than a comparative research investigation.

The key contribution lies in the successful connection of two previously isolated systems—HumanTech and CTD—through data standardization, transformation, and automation using the Palantir Foundry platform. This integration allows ergonomics scores to be visualized alongside critical production metrics such as cycle punctuality, over takt time, and MOST-based timing analysis. The ability to observe these parameters in parallel supports data-driven identification of process inefficiencies related to physical strain or poor workstation design.

The absence of production metrics for stations DAR-FI-10 and DAR-60-80 in the evaluated timeframe suggests that these workstations were either decommissioned or inactive during data collection. Such gaps highlight the importance of aligning ergonomics assessments with up-to-date process status to avoid misleading correlations. Future implementations should include a validation step to confirm station activity prior to overlaying ergonomic scores.

During the implementation, multiple standardization issues were encountered, including inconsistent data entry, missing or incomplete assessments, and deviations from corporate procedures. These findings underscore the need for robust data governance, standard compliance, and continuous monitoring of input quality to maintain the integrity of analytical outputs.

The developed solution also introduces enhanced functionality, such as filtering assessments by product variant, which opens up new opportunities for detailed and flexible analysis. These capabilities are particularly useful for ergonomics teams aiming to investigate localized risks or performance deviations linked to product complexity.

Our findings are consistent with prior work demonstrating the value of automated ergonomics analysis. For instance, Lasota [15] showed that detailed motion capture frameworks can pinpoint high-risk postures, while Bortolini et al. [17] found that AI-driven MAS systems improved both ergonomics and productivity in manufacturing settings. By embedding ergonomics scores directly into production dashboards, our approach extends these concepts toward real-time, shop-floor decision support, thereby bridging the gap between standalone ergonomics tools and process-monitoring applications.

Despite these advancements, several future challenges must be considered. A major concern is maintaining long-term data consistency across departments and regions, which requires ongoing training, user discipline, and possibly the implementation of automated data validation tools. Scalability also presents a challenge, especially as more lines or plants are connected to the platform. Larger data volumes will demand more efficient processing logic and potentially new infrastructure capabilities.

Another emerging challenge is expanding the scope of ergonomic monitoring. While this study focuses on time-based metrics, future iterations may incorporate real-time physiological or sensor-based data, such as motion capture or fatigue indicators. Although promising, this direction raises concerns about data privacy, hardware reliability, and processing complexity.

Lastly, the long-term success of the system depends on effective cross-functional collaboration. Ensuring alignment among EHS specialists, production engineers, IT staff, and operational managers will be essential for maintaining the system’s relevance and maximizing its impact.

This integration effort supports broader Industry 4.0 objectives, where data from different operational domains are interconnected to enhance both productivity and employee well-being. Embedding ergonomics data directly into production dashboards enables not only retrospective analysis, but also real-time insights for daily management. As factories move towards digital twins and holistic monitoring, this approach demonstrates how human-centric metrics can be seamlessly integrated into automated production control systems.

Although the case study describes a single implementation instance, the developed solution is part of a global deployment across more than 300 manufacturing sites within a multinational automotive group. The study captures the early rollout phase, during which foundational data integration and visualization capabilities were introduced. As the application becomes adopted across additional plants, more detailed statistical analyses and benchmarking will be feasible. However, this paper focuses primarily on the technical integration and practical implementation of the solution.

5. Conclusions

This research resulted in the development and deployment of a standardized, automated data integration pipeline for overlaying ergonomics station scores onto production performance charts in the CTD Shift Report application. The implemented solution enables users to visualize the correlation between ergonomic risk factors and production performance metrics in a consistent and scalable way.

By integrating HumanTech and CTD datasets through the Palantir Foundry platform, the project successfully addressed challenges related to data structure differences, inconsistent user input, and system isolation. The use of automated pipelines ensures up-to-date and reliable data flow, supporting both immediate operational decisions and long-term ergonomics management strategies.

The outcomes support the hypothesis that higher ergonomics scores may correlate with lower cycle time punctuality. The improved visibility of this relationship allows production and EHS teams to identify problem areas more effectively and prioritize actions to improve both productivity and operator well-being.

The methodology presented in this study provides a scalable model that can be extended across multiple plants and adapted to include additional health and safety indicators. It sets a precedent for incorporating ergonomics directly into operational control systems, making human factors an integral part of production excellence initiatives.

In conclusion, this approach contributes to the digitalization of ergonomics monitoring in manufacturing and sets a foundation for broader integration of human-centered metrics into operational dashboards.