Headed to #CoRL2026 this week? Check out some of our accepted papers and workshops, covering everything from handheld data collection, loco-manipulation, and robotic perception, to sim-to-real policy transfer and zero-shot long-horizon planning: rai-inst.com/resources/blog…
RAI Institute
Verified account This account is verified. Learn more@rai_inst
We aim to solve the most important and fundamental problems in robotics and AI. (Formerly The AI Institute)
13 Following 8.5K Followers 53 Posts
Latest posts 5
No motion capture. No tethers. Just onboard perception and reinforcement learning taking on the real world. Watch our vision-based policy handle noisy depth streams and intermittent visibility to roll, upright, and re-orient tires in an unmapped skatepark and learn how we built and trained this onboard system: rai-inst.com/resources/blog…
0:30
Robotic manipulation is moving toward adaptive, real-time physical interaction. Building on previous work where a Spot quadruped used whole-body control to manipulate a 15 kg tire, this system now operates autonomously outdoors.
By replacing external motion capture with onboard depth-camera perception and computing locally, Spot can perceive and manipulate objects in unmapped environments. Learn more about how this type of visual policy learning allows the robot to dynamically sequence and adapt behaviors like rolling, uprighting, dragging, and recovering from interference on the fly: rai-inst.com/resources/blog…
By replacing external motion capture with onboard depth-camera perception and computing locally, Spot can perceive and manipulate objects in unmapped environments. Learn more about how this type of visual policy learning allows the robot to dynamically sequence and adapt behaviors like rolling, uprighting, dragging, and recovering from interference on the fly: rai-inst.com/resources/blog…
2:12
In addition to powerful throwing arms, AthenaZero features light, high-speed, tendon-driven hands capable of closing in ~80ms.
Powered by torso motors, they combine high force with extreme delicacy, handling heavy tools or fragile items like eggs. With 11 high-accuracy sensors per fingertip and palm, they deliver precise touch sensitivity and low-latency control.
Learn more about AthenaZero’s tactile sensing and dynamic manipulation capabilities in Science Robotics: science.org/doi/10.1126/sc…
Powered by torso motors, they combine high force with extreme delicacy, handling heavy tools or fragile items like eggs. With 11 high-accuracy sensors per fingertip and palm, they deliver precise touch sensitivity and low-latency control.
Learn more about AthenaZero’s tactile sensing and dynamic manipulation capabilities in Science Robotics: science.org/doi/10.1126/sc…
1:40
While legged robots navigate unstructured environments well, their high energy use and low speeds can hinder longer journeys. Wheeled robots offer efficiency but can struggle with obstacles like curbs or stairs. To bridge this gap between speed, agility, and terrain versatility, we introduced the Ultra Mobility Vehicle (UMV), a single-track bicycle robot that combines wheeled efficiency with dynamic articulation.
Learn more about UMV's system design and how Reinforcement Learning (RL) is applied to control bike jumps, flips, bunny hops, and more in the October 2026 issue of @SciRobotics science.org/eprint/BV4IMXA…
Learn more about UMV's system design and how Reinforcement Learning (RL) is applied to control bike jumps, flips, bunny hops, and more in the October 2026 issue of @SciRobotics science.org/eprint/BV4IMXA…
1:44