Episode Summary:
In motorsports, drifting involves racing on closed courses in a continuous opposite lock slide, balancing the car at the age of control. While it’s not a recommended driving mode for street use, the Toyota Research Institute has developed an autonomous drift car, with a view toward implementing the technology in passenger vehicles. If successful, the system would add a margin of safety by allowing a self driving car to maneuver violently to avoid collisions, while maintaining control beyond the tires’ maximum grip levels.
Access all episodes of This Week in Engineering on engineering.com TV along with all of our other series.
Transcript of this week's show:
To see any graphs, charts, graphics, images, and/or videos to which the transcript may be referring, watch the above video.
Autonomous driving research has for years focused on automation that mimics the actions of a safe human driver. Between the lines, safe distances and within speed limits. But there are circumstances when safety requires operating the car to its limits of adhesion, and perhaps beyond, such as when swerving to avoid an animal, or when freezing temperatures cause black ice. Since the goal of autonomous driving is to keep the vehicle within its normal performance envelope, can self-driving cars adapt to unusual conditions?
Toyota thinks so, and has released a video of what the company calls a new research milestone in autonomy, with a robotic drift car. Drifting is a popular form of motorsport where competitors intentionally break traction to the rear wheels and negotiate a circuit with varying degrees of opposite lock while maintaining vehicle control. Power sliding is not the fastest way around a racetrack, but it does require a high level of car control, and the Toyota Research Institute (TRI) has developed a self-driving prototype that can maintain car control beyond the limit of adhesion in road emergencies.
TRI developed the hardware and software with the Dynamic Design Lab at Stanford University using a race-prepared Toyota Supra drift car and with the support of well-known Drift competitor Ken Gushi at California’s Thunderhill raceway. The Supra was heavily instrumented and was fitted with actuators for throttle, steering and individual wheel braking. Vehicle dynamics information was beamed to an x86 computer from a dual antenna real-time kinetic satellite navigation assisted inertial navigation system at a 250 Hz sample rate.
The algorithms are the key, and TRI has developed a nonlinear model predictive control system to cope with the extra complexity of vehicle operation past the limit of tire adhesion. Model predictive control was developed as a form of process control for factory operations, using dynamic models of processes in real-time, but with a forward-looking capability to predict possible future outcomes. The system must simultaneously optimize vehicle stability using throttle, steering and wheel braking to manage tire slip angles, while processing sensory data about obstacles around the car, anticipating the control inputs that will be necessary to maintain control while taking evasive action.
The problem is significantly more difficult than simple emergency braking or collision avoidance at low speeds. At the current stage of development, the goal is to augment a human driver’s capability by giving the car a very high-performance collision avoidance capability that includes opposite lock slides where necessary to avoid a collision or cope with slippery surfaces in an emergency.
In a sense, the system would give ordinary drivers the skills of a professional racing driver in dangerous situations. Toyota has yet to announce a timeline for availability of this technology in production vehicles.