Introduction
Formula Student competitions push the boundaries of automotive engineering, and the Dynamis PRC team is at the forefront of innovation with its driverless car. The team’s cutting-edge autonomous functionalities are a crucial element in ensuring success on the racetrack. This blog provides a high-level overview of the various modules responsible for the car's autonomous driving capabilities.
1. Sensor Suite
The sensor suite forms the foundation of the car’s perception. It includes:
- Cameras: Two mono cameras are mounted on the main hoop, providing a 50-60 degree field of view each. They work in tandem to detect cones on the track, feeding color data into the system for object detection.
- Lidar: A 10Hz rotational Lidar supplies crucial distance and object detection data with a horizontal FOV of almost 180 degrees. This is mounted in the same location as the cameras.
- Inertial Navigation System (INS): Installed at the vehicle's center of gravity, the INS assists with localization by providing GPS and movement data.
- Optical Speed Sensor: Mounted on the front left wheel, this sensor enhances speed measurement accuracy.
2. Autonomous Steering and Brake Actuators
Autonomous control is executed via:
- Steering Actuator: A BLDC motor drives the steering in a steer-by-wire system, which is critical for precise control during high-speed maneuvers.
- Brake Actuator: The brake system ensures responsive stopping power, maintaining safety during autonomous operations.
3. Autonomous System Computing Unit (ASCU)
The NVIDIA Jetson Orin AGX powers the vehicle's autonomous systems. With its 64-core CPU and AI-focused architecture, it processes sensor data and executes complex decision-making algorithms in real time.
4. Software Framework: ROS2
The software architecture leverages ROS2 for managing sensor communication and processing. Key components include:
- Perception: This module integrates camera and Lidar data to detect track cones and other obstacles.
- Estimation: Using SLAM (Simultaneous Localization and Mapping), the system builds a map of the track while localizing the car. The implementation uses FastSLAM and GraphSLAM to ensure precise positioning.
- Control: The Model Predictive Control (MPC) algorithm computes the necessary steering and braking commands, ensuring the car follows its intended trajectory.
- Interfaces: ROS2 serves as the interface connecting the sensors, actuators, and computing unit.
5. Path Planning and Decision-Making
Path planning relies on:
- SLAM: Localization and map generation allow the vehicle to understand its environment and optimize its path.
- Trajectory Optimization: The car computes optimal paths and velocity profiles in real-time, ensuring safe and efficient driving on the track.
6. Simulator for Development and Testing
To test and fine-tune the system, the team uses a custom simulator based on FSSIM. This allows the team to run simulations of track layouts and analyze vehicle behavior without needing physical access to the racetrack.
7. Safety and Redundancy
The Status Controller oversees the health of all critical systems. It monitors the incoming data and ensures the vehicle can be stopped safely if a malfunction occurs. If a critical message is missed, the car automatically enters Emergency Mode, guaranteeing safety during operations.
Conclusion
The autonomous driving system of Dynamis PRC is a result of careful design, robust software architecture, and powerful hardware integration. The team's dedication to innovation and safety has resulted in a driverless vehicle capable of competing at the highest levels of Formula Student racing.
