On September 17, Boden AI officially open-sourced the RW-RL-HIL-Dataset, a real-world robot reinforcement dataset focused on human intervention data during actual robot deployments. A subset of the RW-RL-Dataset, it totals 82.23 hours across 3,347 episodes and 4.44 million frames, amounting to 188.57GB of data. Covering nine household tasks, it records policy deployment, human takeover, correction, and handover. Data indicates 3,068 episodes feature at least one human intervention, totaling 506,800 frames under human control—11.41% of all frames—with 6,408 continuous intervention segments, averaging 1.91 per episode and a mean duration of 5.27 seconds. Structured in LeRobot v2.1 format, it includes three video streams—head, left wrist, and right wrist—with 14-dimensional state and action vectors, recorded at 640×360 resolution and 15FPS using H264 encoding, with frame-by-frame control flags and intervention boundary annotations. Combined with the main repository, it supports research into human-in-the-loop imitation learning and reinforcement learning. In June, Boden AI teamed up with the JIP Innovation Center and Shanghai Jiao Tong University's MINT Lab to release the initial RW-RL-Dataset, topping 1,000 hours. To date, Boden AI has open-sourced 542 hours—460 hours of teleoperation data and 82