Wednesday, November 18, 2026
24 papers added this day. New papers stay for 14 days.
Tactile sensing 3
Dexterous manipulation 1
Imitation learning and data collection 9
Robot foundation models 4
Legged locomotion 1
Humanoids and mobile manipulation 2
Computer vision for robots 2
Robot hardware and actuators 2
Grana: Compositional Reward Modeling for Long-Horizon Robot Manipulation
November 2026
A vision-language reward model picks the unfinished subtask from a language list, predicts its progress and adds credit for the finished subtasks. Pretrained on 200 tasks, it cuts zero-shot progress error by 20.7% over RoboMeter, and progress-weighted cloning raises real airplane-assembly success from 50% to 75%.
DiWALI: Diffusion Policy Learning in World Models via Latent Action Inference from Human and Robot Play
November 2026
A world model driven by latent actions inferred from pairs of frames learns from robot play and from action-free human play. A diffusion policy imitates human demonstrations in that latent space and improves with RL in imagination. With human demonstrations only, it reaches 59.6% on CALVIN vs 35.2% for the best baseline given five robot demos, and on a real Franka, fine-tuning in imagination raises success from 53% to 70%.
QDD-MM: A Simple and Compact Open-Source Muscle Module with Quasi-Direct-Drive Motor for Tendon-driven Robots
November 2026
An open-source muscle module for tendon-driven robots built around a quasi-direct-drive motor with a 7.75:1 gear, so muscle tension can be set from motor current without a tension sensor. Compared with an earlier 53:1 module it needs about half the force to backdrive (about 4 N vs 8 N) and winds wire about 1.5x faster at comparable tension.
LuxTac: A Flexible, Scalable, and Open-Source Optical Taxel Array with Distributed Three-Axis Force Sensing for Tactile Robot Learning
November 2026
A 3.5 mm thick, flexible tactile array whose taxels are off-the-shelf color light sensors under a silicone gel. A swappable gel with colored walls turns shear into a color change, so each taxel estimates three-axis force (per-taxel R² 0.92-0.94 on held-out sessions). Adding the pad's pressure readings to a visuomotor policy raised peg-in-hole success from 6/15 to 13/15.
TCP4D: A Unified Tool-Center-Point Tracking and Generation Model for Robot Manipulation
November 2026
TCP4D recovers the 6-DoF tool-center-point pose and gripper state of a robot from monocular RGB video, and reuses the same 4D scene features to generate future tool trajectories as a policy. Tracking error is 7.60 mm on DROID, and on two real-robot tasks the policy succeeds 78.33% of the time, against 63.33% for DP3 and 31.67% for ACT.
Levio: Smart Actuated Back-Exoskeleton via Retrofit of a Commercial Passive Device
November 2026
They retrofit a commercial passive back exoskeleton (Laevo Flex) with two off-the-shelf quasi-direct-drive motors, IMUs and a load cell, and switch its assistance between the bending and lifting phases. In 10 kg stoop lifts, it lowered resistive torque while bending from 25.2 to 17.9 Nm and raised lifting assistance from 30.6 to 37.7 Nm compared with the passive device.
VLM-guided Grounding Expert for Generalised Referring Expression Comprehension
November 2026
VGE couples a frozen Qwen3.5-9B vision-language model with a trainable Grounding DINO detector, so a robot can find zero, one or many objects named in a reworded or chatty request. On the robotic-scene RoboRefIt benchmark it keeps 95.2% F1 when context is added to the instruction (Grounding DINO falls to 62.6%) and runs 5.3x faster than a LoRA-tuned Qwen3.5-9B.
Task Convergent Flow Matching for Generative Robotics
November 2026
Before executing each action chunk of a flow-matching policy, they search its noise seed with CMA-ES over simulated rollouts for a chunk whose task error stays inside a funnel that allows temporary setbacks but requires net progress. In simulation, success rises from 31.3% to 88.5% on PushT and from 77.0% to 86.0% on three-cube stacking.
Distilling World Action Models for Efficient Bimanual Manipulation via Anchored Path Matching
November 2026
Distills the LingBot-VA world action model into one student that generates future video and bimanual actions in 1, 2 or 4 denoising steps. One-step APM reaches 80.9% average success on RoboTwin 2.0 and, on a real Astribot S1, cuts control-cycle latency 21.3x versus the eight-step teacher.
InteractVLA: Turning VLAs into Interactive Tools for Agentic Robotic Manipulation
November 2026
InteractVLA lets a vision-language agent steer a frozen flow-based VLA toward an end-effector target pose by guiding its denoising, and read back the planned motion to correct it. It reaches 99.4% on LIBERO-Plus and, on four real bimanual tasks, roughly doubles task scores over a language-only agent on pouring and bottle uprighting.
Heave-Ho: Decentralized Long-Horizon Multi-Humanoid Collaboration
November 2026
Heave-Ho lets teams of Unitree G1 humanoids carry shared objects with no network link between them. Each robot runs one shared residual RL policy on top of a frozen motion tracker, and a leader speaks the next team skill aloud. Real teams succeed in 92 of 100 trials, and in simulation the controller succeeds in 94% of episodes against at most 29% for baselines.
From Temperature-Limited to Battery-Limited Legged Locomotion
November 2026
A reward that penalizes the motor temperatures a gait will eventually settle at, predicted by an identified thermal model, teaches a Unitree Go2 locomotion policy to spread heat without ever observing temperature. Carrying 5 kg at 1 m/s it walks for 174 minutes until the battery is empty, where a learned baseline overheats after 19.
HandUMI: A Modular Hand-Worn Parallel-Jaw Interface for Robot-Free Bimanual Data Capture
November 2026
HandUMI is an open-source hand-worn parallel-jaw gripper, tracked by VR controllers, for recording bimanual demonstrations without a robot; a pair costs USD 557 with a Quest 3S. Trained only on 562 such demonstrations, ACT folds a towel on real PiPER arms in 12 of 20 attempts.
Déjà-Events: Multi-view 3D Geometry with Event Cameras via Sparsity-Aware Adaptive Gating
November 2026
An event-camera-only model that predicts depth, camera poses and pointmaps from several views in one forward pass, distilled from a frozen RGB reconstruction model without any depth or pose labels. Trained only in simulation, it halves the pointmap error of the strongest event baseline on real DSEC driving data (0.15 vs. 0.30 AbsRel) with about a tenth of its parameters.
WinDEX: Metric 3D Hand Motion Reconstruction in the Wild for Dexterous Teleoperation and Policy Learning
November 2026
A head-worn fisheye stereo camera with an IMU and a network trained on 1,000 hours of refined labels recover metric, world-frame 3D hand motion from everyday egocentric video. RL policies that follow these hand trajectories reach 61% simulated grasp–lift–place success, vs 8% and 17% with WiLoR and HaWoR references, and the same pipeline drives live arm–hand teleoperation.
ForceHarness: Turning Force into Action and Physics for Frozen Frontier Models
November 2026
A harness lets frozen, API-only frontier models use force. Self-terminating force primitives run the contact loop under the model, and each force trace comes back as physical quantities such as mass, friction and contact height. On a Franka FR3 that estimates force from joint torques, GPT-6 Astra goes from 2/60 to 54/60 successes on six contact-rich tasks, with no demonstrations or finetuning.
WiTac: Wrist-Informed Tactile Representations for Contact-Rich Manipulation
November 2026
WiTac is a pretrained encoder that fuses each finger's tactile history with wrist-camera views into one contact token per finger, then conditions a VLA policy. On four real bimanual dexterous-hand tasks it raises GR00T N1.7's average success from 37.5% to 66.3%, and π0.5's from 43.8% to 58.8%.
TaF-Foundation: Wrench-Grounded Cross-Sensor Tactile Representations
November 2026
They record over 100 million tactile readings paired with measured force/torque from 24 different tactile sensors and train one encoder that maps all of them into a shared, force-grounded feature space. Against the strongest baseline it cuts single-frame force error by 17.1% and raises average real-robot success on four tasks from 23.75% to 46.25%.
AsymEEF: Bimanual Manipulation with Asymmetric End Effectors
November 2026
AsymEEF maps grippers and dexterous hands into one shared latent state, so bimanual demonstrations teleoperated with one pair of end effectors can be replayed on other pairs and filtered by task success. On six real tasks, 88.7% of transferred demonstrations succeed on replay, against 20.8% and 45.2% for two prior transfer methods.
Lift, See, Act: Hierarchical Robot Policy Pretraining with 3D Foundation Models
November 2026
Pretrains a robot policy's visual encoder on human hand videos by distilling the features and depth maps of 3D foundation models and predicting 3D wrist trajectories. With the encoder frozen, a Franka policy averages 68.3% success across six real-world task settings, against 41.7% for the best baseline.
DexFormer 2.0: Towards a Cross-Embodiment Foundation Policy for Dexterous Grasping
November 2026
DexFormer 2.0 is one RL grasping policy for many arm-hand pairs. Each robot is mapped to shared joint slots described by per-joint morphology, and a Transformer-XL memory supplies recent interaction. Trained on 18 of 24 pairings of four arms and six hands, it reaches 89% stable success on those pairs and 51% on the six held-out pairs in simulation. On a real Flexiv arm, a distilled depth-based student succeeds in 23 of 25 grasps with LEAP and 17 of 25 with Revo3.
Right Goal, Wrong Body: Re-grounding Frozen VLAs under Actuation Shift
November 2026
ReGround keeps a VLA frozen and corrects its actions at test time when the robot's actuation changes (latency, controller gain, action-frame rotation), using a dynamics model fitted on about a minute of motor babbling. On eight simulated LIBERO tasks it raises π0.5's success under 400 ms latency from 0.16 to 0.90.
EgoSwarm: Learning Decentralized Robot Swarm Intelligence from Human Swarm Demonstrations
November 2026
EgoSwarm records people collaborating with head-worn sensors (about 50 hours) and turns the recordings into training data for one shared VLA policy that each robot runs on board, exchanging only short messages with its neighbors. On three real three-robot tasks it reaches 72.2% mean success, against 22.2% when trained without the human data.
MoViC: Real-Time Humanoid Whole-Body Control from Monocular Video Streams
November 2026
MoViC drives a full-size humanoid from a single webcam: the operator is reconstructed at about 10 Hz, a learned extrapolator fills in 50 Hz references, and a transformer tracker conditioned on a compact frequency-domain motion code follows them. The tracker reaches 99% success versus 95% with a larger ten-frame preview, and the real robot carries a box and a bag under webcam teleoperation with no wearables.
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.
