Constructing a realistic virtual training environment can require extensive expert labor before a robot starts learning. Pranjal Aggarwal and Sean Welleck propose Gym-Anything, where coding agents build simulations and an adversarial audit checks whether their physics and tasks satisfy explicit criteria. The selection of physical work draws on US economic activity instead of focusing solely on familiar toy demonstrations.
Read more… →
A conventional room map can show a robot where a door is without revealing which way it moves or how much force opening it requires. Dibyendu Ghosh proposes PhysVLMM, a map that stores probabilistic beliefs about articulated components. After a measured pull, the system updates its estimate, and a planner can ask whether an action is likely to succeed within a stated risk limit.
Read more… →
Human fingers and robotic fingers have different geometry, so copying a hand trajectory does not guarantee the same contact with an object. Haozhe Zhang and collaborators propose TouchBridge, a combined hardware-and-software system with a wearable human demonstration interface and a corresponding robotic interface. Both capture visual and fingertip tactile information to establish correspondence across different hand designs.
Read more… →
An instruction such as clearing a table hides many steps: approach, grasp, move and release. Yet large robot datasets often store only the overall task instruction. Xueyuan Wei and collaborators propose RoboAtom-Data, which assigns consistent time-aligned labels to short, or atomic, actions so learning systems can recover the structure of longer behavior.
Read more… →
Robotic hands can have very different numbers of joints yet face similar problems: hold an object, turn it and coordinate both wrists and fingers. Xiaoyu Yang and Nan Wu propose AcrossDEX 1.0, a pipeline combining human-motion pretraining, simulation-based expert learning, robot demonstration collection and visual feedback during execution.
Read more… →
