Sunday, November 22, 2026
14 papers added this day. New papers stay for 14 days.
Imitation learning and data collection 1
Robot foundation models 4
Legged locomotion 2
3D perception and scene representation 1
Planning and control 1
HRI and teleoperation 2
Assistive robotics and haptics 1
Exoskeletons and wearable robots 1
Robot hardware and actuators 1
PUPPET: Physical-Unified Platform for Presence, Embodiment, and Teleoperation
November 2026
PUPPET lets a remote user drive a real TIAGo mobile manipulator from inside a VR replica of the lab, kept in sync by motion capture, and automatically checks each trial's recorded data. In a 30-person study, VR cut post-learning task time to 68.7 s from 95.0 s with 2D camera teleoperation and raised validation-ready trials from 66.0% to 85.6%.
Teaching Vision-Language-Action Models to Stop Safely
November 2026
SAFI fine-tunes a VLA to stop a prohibited task even when it takes over halfway through. It does so by imitating a "freeze and keep the grip" action at every state of that task's demonstrations while still imitating the robot's other tasks. In LIBERO simulation, prohibited-task completion after takeover falls from 99.7% to 1.3% with no dropped objects, and the other tasks keep 93-97% success.
One View, One Action: Spatially Generalizable VLAs through Observation-Aligned Experts and Cross Graph Consistency
November 2026
OVOA-VLA fine-tunes π0.5 with extra action experts that learn each demonstration in every camera's own frame, while the robot still executes base-frame actions. On a real Franka arm under unseen camera views and shifted start poses, success rises from 28.3% to 58.3% at inference cost comparable to π0.5.
AGE-Capture: Autonomous, General, and Efficient UAV Capture for High-Fidelity Scene Reconstruction
November 2026
AGE-Capture flies a UAV to collect images for Gaussian-splatting reconstruction. It picks views with a lightweight renderability state and stops by itself once further views add little. In ten simulated scenes it beats two online baselines by 3.30-3.53 dB PSNR while using 58.2% of the 900 s budget, and it runs onboard a real quadrotor in nine scenes.
Collision-Free Is Not Enough: Passage-Conditioned Navigation for Wheeled-Legged Robots
November 2026
Store each taught stair and ledge crossing as an event on the route, and make a wheeled-legged robot back up and retry before a ledge unless it is within 10 cm and 10° of the taught entry. Under localization noise this raises in-limit ledge crossings from 24.1% to 88.6%; on the real robot it completed all six runs of a 271 m competition course.
Hindsight Self-Distillation: Teaching Robots to Act from the Present
November 2026
HSD trains a video-backbone robot policy with a teacher branch that also sees later frames of each demonstration, and distills its action predictions and visual summaries into a student that acts from current images only. It reaches 98.50% on LIBERO and 92.31% on RoboTwin 2.0, and raises real bimanual success from 24.17% to 37.50% over Fast-WAM.
A$^2$-VLA: Adapting Vision-Language-Action Policies to New Tools through Agentic Affordance Discovery
November 2026
An LLM agent rebuilds tool-use demonstrations in simulation, works out which part of a tool does the job (a hook, an edge, a support surface), and feeds those features to a π0.5 policy. On a real robot with unseen tools, success rises from 4.7% for plain π0.5 to 35.3%.
Proprioceptive Simultaneous Base and Terrain Elevation Estimation for Legged Robot Locomotion
November 2026
A blind legged robot builds a height map from its own footsteps and feeds it back to its state estimator, like loop closures in SLAM. This stops vertical drift and cuts trajectory error by 22.1% against the best proprioceptive baseline on 296 m of real ANYmal-D data.
Modelling the human-exoskeleton interaction interface - Benchmarking MuJoCo's deformable model
November 2026
The authors benchmark MuJoCo's flexcomp soft-body models against a COMSOL finite-element model on indenting and compressing silicone tissue, including a simplified thigh, as groundwork for simulating the human-exoskeleton interface. MuJoCo follows the force trends (correlation 0.97-0.99) but is stiffer, captures only 4-24% of the finite-element hysteresis, and is faster only in the sphere-indentation case.
Effects of Locomotion and Viewpoint Control on Immersive Robot Telepresence for Navigation and Inspection
November 2026
In a 17-person VR study, people teleoperated a real quadruped through a foam maze to test joystick vs omnidirectional-treadmill driving and joystick vs head-tracked camera control. Head-tracked camera control mattered most. Paired with joystick driving it finished fastest (178 s vs 258 s with joysticks for both) and had the fewest wall contacts (2.8 vs 6.1 per trial).
A Hierarchical Onboard Vision-Language-Navigation Framework for Aerial Robots in Complex Environments
November 2026
VLN-AR runs a fine-tuned vision-language model on a drone's onboard computer to turn language instructions into 3D waypoints, which a classical planner then flies collision-free. It completes 91 of 101 real-world tasks and reaches 78.15% success on the AerialVLN benchmark.
Sideslip-Conditioned Drift Trajectory Planning with Flow Matching and Control Barrier Functions
November 2026
A flow-matching planner generates drift trajectories for a requested peak sideslip angle, control barrier functions correct them for road, friction and actuator limits, and an NMPC tracks them. In the Assetto Corsa simulator the car's median peak sideslip follows requests from 20° to 45° with a mean absolute error of 1.2°.
Haptic Probing for Subsurface-Aware Force Distribution on Layered Soft Tissue
November 2026
Before pressing on soft tissue, a robot hand gives it a brief low-amplitude vibration to tell muscle from bone-backed tissue, then spreads the force over its tactile fingertips under a pressure limit set by that class. Kneading the shoulders of three people with a dual-arm robot, it cut median force-tracking error to 2.22 N from 4.04 N for a fixed-compliance baseline.
An Open-Source da Vinci X/Xi Instrument Module for Laparoscopic Surgery Research
November 2026
An open-source driver mounts da Vinci X/Xi instruments, stapler included, on a UR5 arm, and a relay board lets software fire electrosurgical energy while foot pedals keep priority and every activation is logged by source. Instruments swap in 2.41 s, and the module cut and coagulated ex vivo porcine tissue.
Summaries are written automatically from each paper's text. Figures and videos come from the paper and its project page; all credit goes to the authors.
