A key element of autonomous racing cars is trajectory planning, which determines how the car moves around the track. The success of the race depends to a large extent on how accurately and efficiently the car can follow this predefined line.
What is trajectory and why is it important?
A trajectory is a set of points that define the movement of the car, with each point having a speed value associated with it. This is the ideal trajectory that the car will try to follow, although in reality there will always be minor deviations. The trajectory is influenced by many factors, such as the physical limitations of the vehicle, race strategy and the race rules.
A trajectory is a set of points that define the movement of the car, with each point having a speed value associated with it. This is the ideal trajectory that the car will try to follow, although in reality there will always be minor deviations. The trajectory is influenced by many factors, such as the physical limitations of the vehicle, race strategy and the race rules.
What does a ’good' trajectory look like?
A well-designed trajectory takes into account:
- The fastest possible lap time, with maximum use of the ideal arc and grip.
- The car's capabilities, such as cornering speed and stability.
- The track limits, as the rules may prohibit crossing white lines.
- The accuracy of the sensors, so that the car can accurately determine its own position, taking these into account to define the tolerance bands.
There are also software aspects to trajectory planning. For example, it is important that the curvature changes are as smooth as possible, because this helps the controller to manage the car's movements more accurately.
For tests and competition, different global offline trajectories are needed, such as overtaking, where you need to investigate where overtaking is possible and to do this you need to create several possible trajectories. A similar special case is the trajectory out of the pit and through the pit lane, where the car has to go very slowly but very accurately.
The decision-making process - how does planning work?
The first step in planning is to load the pre-made global offline trajectories. When the car is running it will search itself on the global trajectory and select a local section, this local section can be selected based on the speed of the car, it is not a fixed section, it is time parameterized. And finally this local section is sent to the controller, which controls the movement of the car.
The global trajectory is the reference on which the car chooses the local curve, but when overtaking or when exiting the pit and switching to another global trajectory, it also plans a local transfer curve so that the switch between the two trajectories is based on a nice continuous planned curve.
Planning the right trajectories is essential for autonomous racing cars. Speed, stability and compliance with the rules all depend on it. The combination of online and offline planning strategies allows the car to make the best possible decision in every situation on the track, maximising its performance and competitiveness.
The blue and yellow arcs in the video below show the edge of the track that the car must not touch, while the arc in between is the trajectory that the car will follow.
