Robotics
Research on Robotics: explanations, evidence, limitations and connections between papers.
30 results
ASTEROID teaches partial-view agents to explore from experience
ASTEROID teaches exploration from student histories and used fewer training steps than reinforcement-learning methods in simulated manipulation.
A robot safety filter learns to preserve recovery options
SAFTER filters quadruped commands using predicted recovery; tests show gains on some shifted maps, with limits on the hardest shift.
A driving model can see a hazard yet fail to brake
A simulated driving study finds that direct stopping guidance can make a frozen model avoid obstacles in tested braking scenes.
Simulation-trained robot navigation improves on the real iw.hub
On a real iw.hub, navigation trials required arrival within 0.5 m at speeds below 0.1 m/s and 0.1 rad/s, without collision or timeout.
One learned model helps humanoid controllers scale across motions
SuperMimic reaches SONIC’s released score on a shared motion test with far fewer GPU-hours, despite different training setups.
Robots can choose fresher action chunks during inference
PACT-VLA uses newer observations to select robot action chunks, with higher average success in tested delayed simulation and robot tasks.
A robot world model predicts touch and task outcomes
PhysCanvas predicts robot touch and task outcomes; on four tasks, added state guidance improved agreement with real success rates.
A driving model responds better when it learns alternate paths
Cartender trains a driving policy on alternative paths for the same scene and improves paired instruction fulfillment on fixed recorded inputs.
Robot fleets improve a shared policy from task experience
Robot fleets used failures to select successful actions, improving Phone Packaging on two held-out robots after initialization.
Training robot policies to act after they move the camera
ActiveAlign adapts robot training to camera-moved views, improving task success in perturbed-start simulation and physical occlusion trials.
A smoother keeps violating samples in the trajectory update
A trajectory method keeps smoothed safety estimates below exact values and retains violation gradients in robot simulations.
Robot touch training improves control with and without touch
Spark-0 uses tactile pretraining to improve vision-only robot manipulation scores on the tested Ego Humanoid tasks.
A driving test’s comfort score can reward staying still
A corrected simulator comfort score stops tracking time spent stopped across eight models, changing how driving comfort is assessed.
Robot alarms can predict failure yet still raise task costs
RESCUE tests robot alarms with stop-and-reset; evaluated monitors did not lower task-completion cost versus no intervention on primary LIBERO-Plus tasks.
A mapless depth-image query speeds reactive drone collision checks
A GPU method queries depth images directly, reducing collision-check latency in a tested 400-query drone-planning setting.
One robot model learns to predict scenes and control varied robots
GreenWAM’s robot-specialized starting point improves RoboTwin task success over its parent under a fixed adaptation recipe.
Match future predictions to the decision they inform
A single image may show an object’s position without revealing its motion. A driving planner must compare possible paths, while a model trained on robot actions may predict success for an action that fails. Each decision calls for a different use of future evidence. The central question is where evidence enters a decision and what the evaluation can establish. The reported findings answer it within different tasks and protocols; they do not establish a shared gain in robot control.
A driving planner organizes candidate paths by motion
ManeuverFlow groups driving paths by motion and uses shared future-scene evidence to score and refine them.
Robot models improve when actions explain intended effects
Arc adds causal traces to existing robot demonstrations and improves control across tested simulation and hardware tasks.
World-model rehearsal helps robots retain skills while learning
D2R helps robots learn sequential tasks while retaining earlier skills through regenerated world-model rehearsal.
Reflex helps a humanoid catch thrown boxes with moving-camera vision
Reflex separates body control, box-motion prediction, and vision so a humanoid can catch thrown boxes, with long-range limits.
Robots can fail during learning, planning, or instruction following
A robot can falter after learning another task, choose poorly among driving paths, or respond differently to two phrasings of a request. These failures arise at different decisions, so the evidence that might help enters at different points. The studies suggest rehearsing earlier tasks during adaptation, consulting predicted scenes while selecting a path, or rewriting an instruction before action. Rehearsal reduces stored trajectory memory while using model computation; the driving planner shares forecast work across paths. Each reported gain belongs to its own test.
GLIO2 completes a route with weak LiDAR geometry
GLIO2 jointly updates LiDAR, inertial, and satellite measurements, completing a tested degenerate bridge route where LiDAR systems diverged.
Human motion helps humanoid robots follow instructions
VioLA uses shared human and robot motion targets to train humanoid control, with strong results in one laboratory evaluation.