Sunday, November 15, 2026
20 papers added this day. New papers stay for 14 days.
Robot foundation models 1
RL and sim-to-real 3
Humanoids and mobile manipulation 1
Computer vision for robots 4
3D perception and scene representation 2
Planning and control 7
HRI and teleoperation 1
Assistive robotics and haptics 1
Heavy Metal Acrobatics: Learning Dynamic Control for Skid-Steer Loaders
November 2026
RL policies trained in simulation, with a neural model of the hydraulic drive fitted to real driving data, make a 5-tonne skid-steer loader swing up onto its front wheels, balance, drive, spin and land softly. Trained only on flat ground, it drives over 75 cm bumps that stop an MPC baseline.
Natural Swing-Up of an Underpowered Double Pendulum
November 2026
A gain-free controller swings up a direct-drive two-link robot whose motors give only 6% of its shoulder gravity torque. It splits torque into work-free steering and energy pumping, chosen by a two-variable linear program. On hardware it reaches near-upright energy in 13.6 s and 7.7 s along the robot's two nonlinear normal modes, staying closer to each mode than two baselines.
An Integrated Architecture for Data-Efficient Real-Time Neural Imitation of Nonlinear Time-Optimal Model Predictive Control
November 2026
Imitate a slow time-optimal MPC for swinging up a Furuta pendulum with a small neural network, querying the expert only where the network's own predicted time-to-go stops falling as it should. This matches DAgger with less than half the data in about a third of the training time, and the network runs at 250 Hz on hardware where the MPC manages 1.36 Hz.
BlinkMimic: Training a Humanoid Motion Tracker in Minutes on a Single GPU
November 2026
A training recipe that fits one whole-body tracking policy for all 40 LAFAN1 motion clips on a simulated Unitree G1 in 3.4 minutes on a single RTX 4090, against 7.4 minutes for the InstantTrack baseline re-run on the same GPU.
Alloy: A Multi-Material Physics Engine for Robot Learning
November 2026
Alloy is an open-source engine that puts rigid robots, cloth, cables, liquids, granular material and soft bodies into one scene. It routes each object to a suitable solver and couples the solvers inside every control step, across many parallel worlds on one GPU. Coding agents that had never seen it trained policies for quadruped walking (97% held-out success), cable lifting (95.7%) and pouring (92%) on a single L4 GPU.
Robot Navigation and Obstacle Avoidance from Randomly Positioned External Cameras
November 2026
VINE lets a robot navigate a room from randomly placed external cameras by predicting a top-down obstacle map from their images and handing it to a PPO policy or an A* planner. In simulation it reaches 92-95% success, above PPO with an onboard camera (83.9%).
Recover, Don't Just Continue: Learning from Feedback for Physical Failure Understanding and Correction
November 2026
A multimodal "corrector" model, trained on failures injected into successful simulated trajectories, checks each finished subtask, names the observable cause of a failure and writes a recovery instruction for a frozen executor. On a real Piper arm it raises macro task success from 45% to 62% for π0.5 and from 51% to 70% for LingBot-VLA.
Single-View RGB-D Force--Deformation Estimation for Robotic Manipulation of Elasto-Plastic Materials
November 2026
A Franka FR3 presses into Play-Doh while a wrist-mounted RGB-D camera measures the dent it leaves. A small MLP maps the estimated dent volume back to the robot's own force estimate with a held-out RMSE of 1.857 N (R² = 0.440).
Reading the Table: Sim-to-Real Game State Detection for the Chef's Hat Card Game
November 2026
Render 25,413 auto-labeled, domain-randomized images of the Chef's Hat card game in simulation and finetune YOLO detection and segmentation models on them. On real images from a NICO humanoid's camera, the pipeline assigns cards to game zones with an F1-score of 0.906, against 0.627 without randomization.
Elastra: A Simulation Framework for Elastic Bodies
November 2026
Elastra simulates mattresses and trampolines in MuJoCo as grids of cells that sink and rebound under contact, checked against an independent implementation and a real mattress load curve. A residual RL policy trained in it raises a simulated Unitree G1's get-up success with HoST from 44.1% to 72.5% overall, and from 55% to 89% on a trampoline.
Decide Before You Move: Diffusion-Generated Search Reference for Zero-Shot Open-Vocabulary Object Navigation in Semi-Static Environments
November 2026
PreNav plans an object search before the robot moves. An LLM scores which furniture the target is likely to be near, and a diffusion model turns that score and the wall map into one ordered search path. It beats SGM on Gibson ObjectNav when both get the same priors, matches its TSP teacher while generating paths 25.8x faster, and finds open-vocabulary targets in 13 of 15 trials on a real Fetch robot.
MiniPose: Miniature-Based Real Data Collection for 6D Pose Estimation of Large Industrial Pipe Segments
November 2026
Collect real RGB-D training data for 6D pose estimation of large industrial pipes by capturing a 3D-printed half-scale replica and rescaling its depth. Adding 1,000 miniature images per pipe type to 10,000 synthetic ones raises mean ADD-S AUC on real full-scale pipes from 91.9 to 94.5 for PVN3D and from 93.5 to 94.5 for RDPN6D.
Open-Source Transtibial Prosthetic Bypass Preserving Joint Alignment: Gait Learning and Kinematic Evaluation
November 2026
An open-source transtibial prosthetic bypass (2.2 kg, about 1000 € in parts) lets able-bodied people walk on a prosthetic foot without a shoe lift or a sideways offset. Six first-time users walked on a treadmill without handrails after about 10 min, with prosthesis-like ankle motion but swing knee flexion cut from 61° to 36°.
SIGMO: Sigmoid Gated Interaction-Aware Motion Planning
November 2026
SIGMO estimates how likely each nearby car or pedestrian is to yield from five hand-designed sigmoid features, and uses that estimate to scale each agent's risk inside an MPPI planner. In a packed-merge simulation it raises success from 26% to 38% over plain MPPI, and it runs at 15 Hz on a 1/10-scale F1Tenth car.
Gaussian Process-Augmented Model Predictive Contouring Control for 6-DoF Trajectory Tracking with a Holonomic AUV
November 2026
A 6-DoF model predictive contouring controller for a small holonomic underwater vehicle, with Gaussian-process residuals that correct the rotational dynamics. In a test tank it cuts the attitude error of the vehicle's field-proven PID controller by 47% on a lawnmower survey, and it follows an inspection spiral that the PID cannot.
Optimization-Based Path Planning for Industrial Manipulators: Experimental Validation in Additive Manufacturing
November 2026
An offline planner optimizes a UR10e's whole joint path to meet process tolerances in carbon-fiber winding and 3D printing, and an iterative scheme sets its cost weights from KPI gaps. The planned winding path's worst fiber-alignment error is 1.90°, under a 2° limit, and the 479,523-pose printing path stays within 0.066 mm of the reference.
Rhythmic-Semantic Co-Speech Gestures for Emotional Dialogue in Robots
November 2026
A text-only system that gives the social robot Haru both iconic, emotion-carrying gestures and rhythmic beat gestures while it speaks, using small BERT-style models trained on BEAT2. In an online study with 60 participants it scored highest on speech-gesture correlation and was ranked best most often (40% of trials).
Incorporating World Models in Decision Making for Next Best View Selection in Active Sensing
November 2026
A small Gaussian-process adapter learns where a structured-light sensor returns no depth, so next-best-view selection with a frozen pretrained world model stops picking views that come back empty. On a held-out reflective helmet it reaches 98.4% surface coverage in 4 captures; nominal selection is still at 97.5% after 10.
SAM-VTP: Video Memory and Measured Geometry for Continuous Traversability Segmentation
November 2026
SAM-VTP fine-tunes SAM2's video memory for traversability segmentation, adds slope, roughness and height measured from depth, and trains on continuous rather than binary targets. On a physical robot, the planner found no feasible path 3.3x less often than with a frame-wise SAM2 baseline on a paved course and 6.9x less often on rough ground.
Learning Contextual Motion Constraints for Object Rearrangement Exploration
November 2026
From one partial point cloud, MoCo-Net predicts how far an articulated part such as a drawer or door can move before it hits something, and what it would hit. In 1,000 simulated cluttered scenes, a planner using both predictions reaches the target part state in 82.4% of tasks with 0.201 collisions per task, versus 0% and 2.525 when moving the part directly.
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.
