Leveraging real-time data visualization, fast data extraction, and AI-powered reporting for safer, smarter autonomous vehicle trials.
For over a decade, the Renewable Energy Vehicle (REV) Lab team at the University of Western Australia has been pioneering electric and autonomous vehicle technology. Since their founding in 2008, they’ve led the way in sustainable transportation solutions. At a time when commercial electric vehicle companies were just beginning, they were already designing and building Western Australia’s first electric vehicles and charging stations—a significant achievement for the region.
In 2023, after two years of on-campus autonomous driving trials, the team achieved a critical milestone: securing approval from the WA Department of Transport for public road testing. Supported by DyFlex Solutions, C.D. Dodd Metal Recycling, and UWA’s School of Engineering and Business School, the team launched autonomous shuttle trials in Eglinton. The shuttles traverse a challenging single-lane route from office buildings to the beach, providing an ideal environment for real-world autonomous testing. This scenic path combines complex navigation scenarios with varying terrain, enabling the team to validate and refine their systems under real operational conditions.
Ensuring smooth shuttle performance in dynamic road conditions demands precise data analysis. Enter Foxglove—a multimodal data platform designed to visualize and debug autonomous systems in real time. “With Foxglove, the team rapidly diagnoses performance issues, detects disengagements, and resolves them efficiently” says Xiangrui Kong of the REV Lab team. In combination with MCAP, which provides a standardized format for multimodal logging and enables fast data extraction, Foxglove enables the team to manage and analyze large volumes of complex data more efficiently than ever.
The team’s custom Foxglove layout used for data visualization and analysis integrates maps, lidar feeds, camera views, and system status indicators into a unified view. This comprehensive data visualization simplifies diagnostics, enabling the team to pinpoint anomalies and refine system behavior, ensuring safe and efficient shuttle operations. As shown in Figure 3, this comprehensive view supports precise diagnostics, enabling the team to enhance the shuttle’s performance and reliability on future trips.
Managing vast datasets generated during autonomous trials can be complex. MCAP, a high-performance data storage format, simplifies this process by enabling fast data extraction. The team leverages MCAP’s compatibility with ROS environments to extract critical metrics from rosbags.
During disengagements, the team’s researchers pull the shuttle over, extract relevant data using MCAP, and analyze it through Foxglove. The team uses a simple Python script to automate the data retrieval, streamlining their debugging process. This approach minimizes downtime and accelerates performance improvements.
Generating detailed incident and progress reports for autonomous vehicle trials is a regulatory requirement for Australian roads. For each trial, the team must submit both regular progress reports and specific incident reports to traffic managers and regulatory agencies. These reports help authorities monitor progress and gain insights into the challenges and successes of the team’s autonomous shuttle trials. However, processing tens of thousands of data points per trip can be time-consuming—making report generation a daunting task. To tackle this, the team combines MCAP’s data extraction with generative AI models to automate report generation.
The automated reporting system has transformed the way they handle data. Each time the team completes a road trial, they simply run a Python script that pulls relevant data from the MCAP files. This data is then fed into a generative AI model, which compiles the information into a structured report. This AI-driven approach allows the team to generate both regular and incident reports swiftly and accurately, reducing the workload on the team and ensuring they meet regulatory requirements without delays.
As the team looks to the future, Foxglove and MCAP will continue to play integral roles in their autonomous vehicle trials. Foxglove’s intuitive layout and real-time incident capture capabilities give the team unparalleled insight into the shuttle’s performance, while MCAP enables rapid data extraction and analysis. Together, they allow the team to conduct more efficient trials, respond to issues promptly, and ensure that their shuttles operate safely and smoothly on public roads.
By transforming raw sensor data into actionable insights, the REV Lab is pushing the boundaries of autonomous technology. Their work not only advances electric and autonomous vehicles but also sets a new standard for public road testing in Western Australia. Stay tuned as they continue to redefine sustainable transportation through cutting-edge innovation.
To learn more about The REV Lab’s work check out their website at https://roblab.org/.
A special thanks to Xiangrui Kong for helping put this article together.
