Toyota Unveils Basketball-Shooting Robot With Vision Tech
The 74kg humanoid uses reinforcement learning and sensors for near-perfect shooting precision.
Toyota's Frontier Research Center has introduced CUE7, a significantly upgraded basketball-shooting humanoid robot designed as a testbed for advanced robotics technologies. The robot executes a precise routine: it visually locks onto a hoop using sensors to gauge distance, makes micro-adjustments to its arm angle and posture, and releases the ball with carefully controlled force and trajectory. This isn't just for show; the controlled, repeatable task of shooting a basketball allows Toyota to rigorously test integrated systems in computer vision, real-time sensing, motion planning, and embodied AI in a dynamic physical environment.
Substantial engineering improvements define the CUE7. Toyota cut its weight by over 38%, from 120kg to 74kg, and implemented an inverted two-wheel structure for better mobility. The core advancement is a new hybrid control system that fuses reinforcement learning—where AI learns through trial and error—with model predictive control for planning movements. This AI-driven setup aims to enable more agile and dynamic on-court motion. The project builds on a proven lineage; its predecessor, CUE6, improved ball tracking with redesigned hands and foot cameras, leading CUE7 to secure a Guinness World Record in December 2024 for the farthest basketball shot by a humanoid robot at 24.55 meters.
- CUE7 weighs 74kg, a 38% reduction from the previous 120kg model, enabling a lighter, more agile build.
- It uses a hybrid AI control system combining reinforcement learning and model predictive control for dynamic movement and precision.
- The robot lineage holds a Guinness World Record for the farthest humanoid basketball shot at 24.55 meters (over 80 feet).
Why It Matters
This public demo showcases tangible progress in integrating AI vision, sensing, and motion control for real-world physical robotics applications beyond the lab.