The engineers working on autonomous driving need a vehicle model that represents the same dynamics characteristics as the actual vehicle.
When we train the AI controller to drive the actual vehicle, the vehicle model needs to incorporate not only the correct mass and engine power, but also other correct behaviors like braking efficiency, or the load transfer during cornering events. All these performances are heavily influenced by the fundamental suspension designs (dampers, antiroll bars…) and the tire-road interactions.
Furthermore, depending on the scenario that the simulation needs to address, having vehicle models with different level of complexity can be handy. For example, for a common scenario such as emergency braking on a highway, a simplified model is preferred so a higher number of scenario permutations can be verified in a given amount of time. For a more dynamic scenario that perhaps involves a swift lane change to avoid a crash, a higher fidelity Adams Car model with a well-correlated suspension system is going to be essential.
Not only does VTD provide its own vehicle dynamics model, but it also seamlessly connects with the industry’s best-in-class vehicle dynamics simulation tool Adams. Integrating a high-accuracy vehicle model with a realistic 3D environment enables engineers to evaluate complex self-driving scenarios with confidence.