We are developing technologies to amplify drivers by using interactions to instruct and cue the driver for a better and safer experience.

Our research involves changing the way the car interacts with the driver’s steering to improve car-driver team results. For example, we use reinforcement learning approaches to have the car and the driver jointly decide and execute high-level maneuvers such as overtaking.

Our team develops new approaches to improve driver performance, using insights from machine learning, human-computer interaction, controls, hardware, and systems engineering.

Our solutions include interactive prototypes that provide AI-based performance-driving coaching and assistance in the form of verbal feedback, shared control, XR visualizations, and performance data visualizations. 

Our customers include new and existing members of the performance-driving community and everyday drivers who might seek to benefit from AI-based driving coaching.

The challenges include being able to provide both real-time and post-drive feedback that is personalized to an individual’s specific skill and learning needs, using the right communication modality and density of information to balance the challenge and fun of learning.

We are developing human-machine interactions that improve track driver skills without a human instructor or overloading the driver. This includes a dashboard that allows drivers and instructors to review key driving concepts and metrics to provide feedback for the next trials. We seek to develop real-time augmented reality overlays, audio cues, and shared control methods to future scaffold learning in extreme driving conditions. 

We are developing computational models that provide novel ways to help make decisions that result in better driving performance, while retaining, to the extent possible, their original intent while keeping them safe.

Our work in shared autonomy includes Driving Sensei, a manifestation of the combination of the research pillars to produce a driver experience like that experienced by a master teaching a student to reach excellence. We model Driving Sensei after real instructors teaching performance driving skills, and test novel approaches in AI toward better coaching of drivers throughout their experience on and off the track. 

• Personalizing driver safety interfaces via driver cognitive factors inference
• HMIway-env: A Framework for Simulating Behaviors and Preferences to Support Human-AI Teaming in Driving
• Dreaming to Assist: Learning to Align with Human Objectives for Shared Control in High-Speed Racing
• Computational Teaching for Driving via Multi-Task Imitation Learning
• Driving Sensei: Unlocking Driving Mastery with AI
• Gliding on Simulated Ice: Effect of Low-μ Emulation on Drift Training