Autonomous Vehicle Development

Embed has been intimately involved in Autonomous Vehicle Development, participating in UK-Autodrive, the UK’s largest Self-Driving Vehicle research project.

Embed worked with Tata Motors European Technical Centre on the development of an autonomous vehicle. The project was led by Jonathan Clark of TMETC and the base vehicle is a Tata Hexa, Tata Motors’ flagship SUV.

The computing framework chosen was the Robot Operating System. ROS was used to plumb together the off-the-shelf LIDAR and computer vision object detection systems and a bespoke trajectory and motion control system. Together, these deliver completely autonomous driving. Embed aided in the setup of the ROS environment and created the necessary modifications to the Linux kernel for it to run.

As part of the underlying platform, Embed created a custom Watchdog to oversee all the autonomous systems and flag a warning to the human driver should any readings become out of bounds. It also integrated with the behaviour planner.

A novel trajectory planner utilizing model predictive control was developed under a PhD research program to plan the vehicle manoeuvres, these fed into the motion controller which in turn controlled the vehicle platform. Embed aided in the development of the motion controller and were heavily involved in the integration of the behaviour planner, trajectory planner and motion control algorithms into both the vehicle and simulation platforms.

The ROS platform integrated with off-board computer simulations to provide a testing and development environment. These simulations created entire road networks and test tracks with simulated vehicles to mimic real test conditions so that changes to the platform, algorithms and controllers could be assessed without having to perform real-world tests.

To aid in understanding the performance of the system, Embed produced several post-processing tools to provide metrics on the sensor object detection and vehicle control systems. This software gives an objective measurement on how calibration changes affect the performance and suitability of each system.

Embed aided in the development of the V2V and V2X communication systems for the vehicle which provided communication between vehicles (such as emergency vehicles) and infrastructure (such as traffic lights) to enhance the information available to the vehicle for decision making.

Throughout the project Embed utilised a variety of technologies and tools, including; Linux, Java, C, C++, Python, PyQt, Gtk+, Simulink, MATLAB and ROS.

Please check out Dr Mark Tucker from Tata Motors talk, where you will learn what it takes to develop the complex control systems required for an autonomous vehicle.