Fresh juice


Toyota Research Institute emphasizes safety of AI-assisted driving

While recent autonomous vehicle mishaps have provoked safety concerns, Toyota Research Institute (TRI) offered a glimpse into its unique human-centered approach aimed at near-term driver assistance over full automation. In demonstrations at Thunderhill Raceway, TRI displayed research toward AI that enhances drivers rather than replaces them.

TRI's Human Interactive Driving division develops technologies focused on augmented vehicle operation versus robotaxis. "Can we bring automated assistance to more people sooner to save lives?" posed director Avinash Balachandran. "By rethinking human-AI collaboration, we can create safer and more engaging experiences."

The team's pillars span human behavior modeling, expert skill training for AIs using simulation and real-world data, and shared autonomy where drivers retain control in harmony with intelligent aides. Systems provide audio, visual and haptic guidance to keep humans attentive and informed.

In demos, journalists witnessed AIs mastering self-drifting, evasive maneuvers, and even natural language driving instruction through voice interaction. TRI's Driving Sensei concept envisions AI coaches continually tailoring feedback to boost human capabilities over time.

Balachandran summarized, "Teaching expert skills to AI while keeping drivers confidently engaged delivers safety and enjoyment. We're discovering better ways for intelligent systems to partner with people."

TRI's stepwise direction stands apart from full autonomy pushes by competitors. CEO Gill Pratt underscored their commitment to human augmentation: "Our priority is discovering how humans and AI can collaborate even better - amplifying people by predicting actions and enhancing performance."

The institute believes gradually layering assistance tuned to complement drivers can penetrate the market faster to prevent accidents. Blending simplified autonomy with human oversight could thus advance public adoption and trust in vehicle automation. TRI's nuanced approach ultimately promises to transcend either humans or machines alone.

Share with friends:

Write and read comments can only authorized users