Robotics & embodied AI

Robot learning, manipulation, navigation, vision-language-action models, embodied agents in the physical world

41 articles

All Agents Language models Reinforcement learning Vision Generative media Speech & audio Robotics & embodied AI Safety & security Evaluation & benchmarks Efficiency & systems Code & software AI for science & health Theory

Viewgrid list compact

Replay-free continual learning method prevents forgetting in vision-language-action models

The authors introduce SAMBAR, a continual learning algorithm for vision-language-action (VLA) models that requires no replay of previously seen demonstrations. SAMBAR frames continual learning as a constrained optimization problem, solved via the method of multipliers, and selectively anchors parameters important to earlier tasks. On the LIBERO benchmark and in hardware experiments, SAMBAR retained every learned task, while all replay-free baselines fully forgot the first task they were taught.

3 Oct 2026

Robotics & embodied AI

Compressed world models make zero-shot driving plans far faster

Zhu and colleagues evaluate action-conditioned joint-embedding predictive architectures, or JEPAs, as driving world models without training a driving policy. Their proposed AD-E2E-JEPA compresses the latent patches used for planning and reports a 100-fold inference speedup, while retaining planning accuracy and improving representations transferred to imitation learning.

3 Oct 2026

Robotics & embodied AI

A single world model enables robots to insert unseen parts zero-shot

Hansen et al. present InsertionWM, a framework that trains a single world model on up to 90 geometrically diverse insertion tasks using wrist-mounted camera images and robot proprioception. The model achieves 56% zero-shot success on unseen objects with unknown geometry, versus 7% for a model-free baseline. Finetuning the generalist model on held-out objects improves data-efficiency and asymptotic performance.

3 Oct 2026

Robotics & embodied AI

Canonical gripper-frame representation achieves zero-shot cross-embodiment transfer for two-finger manipulation

The authors propose an interaction-centric framework that uses a parameterized universal gripper abstraction to transform RGB-D observations into a canonical gripper frame. This enables zero-shot cross-embodiment and cross-viewpoint transfer for two-finger gripper manipulation tasks, demonstrated in both simulation and real-world experiments. The approach simultaneously achieves competitive benchmark performance and extreme generalization to heterogeneous robot platforms.

3 Oct 2026

Robotics & embodied AI

Trajectory optimization that preserves endpoints improves closed-loop driving

Zhang et al. introduce Endpoint-Constrained Optimization (ECO), a training-free post-processor for end-to-end driving models that fixes the vehicle's executed history and the policy's predicted endpoint, then optimises the intermediate waypoints for physical plausibility. Across two closed-loop simulators and five datasets, ECO improved the aggregate score of every evaluated policy, with gains up to 71% in HUGSIM score for the VaVAM model.

3 Oct 2026

Robotics & embodied AI

Aiming at observed intermediate targets beats final-goal scoring in frozen world model planning

The authors show that scoring predicted outcomes by their distance to the final goal can break latent world model planning even when dynamics are exact and short-horizon search is globally optimal, whenever the route to the goal initially moves away. They introduce Anchored Planning, a training-free method that reuses the frozen model's own offline trajectories to select an intermediate target. Across Cube, PushT, Reacher, and TwoRoom, it outperforms the LeWM planner and additional final-goal search on long-range goals.

3 Oct 2026

Robotics & embodied AI

Semantic messaging helps distributed robots coordinate long manipulation tasks

Zhou and colleagues present DuoMind, a hierarchical system in which each robot independently plans, communicates with peers and executes detailed actions. Across the new RoboPoly benchmark and the existing RoboTwin benchmark, the authors report that this combination improves coordinated task performance, while ablations indicate that both semantic communication and hierarchical control contribute to the gains.

2 Oct 2026

Robotics & embodied AI

Robots improve reusable manipulation skills through guided simulation practice

Wang and colleagues present Reconstruct, Practice, Go Real, a framework that improves a robot execution system through simulated practice without changing the underlying model weights. Across 22 manipulation tasks, success on held-out initializations rose from 28.6% after one practice round to 95.0% after 15 rounds, while a frozen version completed 30 of 30 reported physical trials across three tasks.

2 Oct 2026

Robotics & embodied AI

Simulation-first framework lets coding agents learn real robot skills in 10 minutes

The authors introduce SimEX, a two-stage autoresearch framework that tightly integrates simulation with LLM-based coding agents. In the first stage, the agent conducts open-ended probe-and-optimize iterations in simulation to develop a robust robot toolbox; in the second stage, the toolbox is adapted to a real robot through only a few physical trials, each correcting the simulator and using the corrected simulator to diagnose failures and screen repairs. On three challenging real-world manipulation tasks—towel folding, barcode scanning, and plate manipulation—SimEX achieved success without any demonstrations and with only 10 minutes of real-robot interaction.

1 Oct 2026

Robotics & embodied AI

Robot agents improve by building and reusing executable motion tools

Ge et al. introduce the Universal Robot-Agent Interface, or URAI, which divides robot control between executable code and a frontier-model agent. Code performs complete multi-stage motions, while the model examines feedback and chooses the next tool. Across five simulated manipulation tasks, this structure increased aggregate success from 18.0% with direct fingertip control to 53.0%.

1 Oct 2026

Robotics & embodied AI

Competitive self-play discovers compact motor skills humans can directly control

Rho et al. introduce Game-Guided Skill Discovery, which trains embodied agents through competitive self-play while constraining their high-level control to a small set of discrete skills. Across ant, robot-arm and humanoid environments, the resulting behaviors were directly usable by humans and could be composed into more complex actions, including solutions to previously unseen tasks.

1 Oct 2026

Robotics & embodied AI

Touch-directed curiosity helps robots discover useful manipulation skills

Iten and colleagues introduce TacEx, an exploration framework that rewards robots for reducing uncertainty specifically about tactile outcomes rather than about every unpredictable transition. The authors report that this focus drives contact-rich exploration without task rewards or expert demonstrations, yielding data that supports offline pick-and-place learning and sample-efficient post-training of vision-language-action models.

1 Oct 2026

Robotics & embodied AI

Dense subtask labels improve robots’ long-horizon bimanual manipulation

Choghari et al. introduce FineART, a dataset that pairs long, two-handed robot demonstrations with language labels for each intermediate step. Mid-training a vision-language-action policy on these annotations produced large gains on spatially ambiguous and previously unseen tasks, while reducing the demonstrations needed to adapt to new hardware.

30 Sept 2026

Robotics & embodied AI

Robots recover more often when vision-language models monitor failures

Li and colleagues test whether robot policies can recognize and correct failures during manipulation, finding that retries, textual error descriptions and best-of-N selection rarely produce reliable recovery. Their Spotter system instead assigns reflection to a vision-language model (VLM), improving simulated and real-robot success rates while avoiding the latency of VLM-led control.

30 Sept 2026

Robotics & embodied AI

Structured intermediate rewards improve robots on long, dependent tasks

Yin et al. introduce StructRL, an online reinforcement learning framework for vision-language-action models performing extended sequences of dependent robot manipulations. Across two simulation benchmarks and two foundation policies, the authors report that structured, pace-sensitive subtask rewards outperform the evaluated online RL baselines that rely more heavily on terminal success.

30 Sept 2026

Robotics & embodied AI

Counterfactual videos teach humanoids to handle unfamiliar objects

The authors present PRISM, a real-to-sim-to-real pipeline for training humanoid robots from a small number of human interaction videos. By generating varied counterfactual videos and grounding their reconstruction in human-object contact, they train one depth-based policy that picks up, carries, and drops novel objects on a real robot without real-world fine-tuning.

30 Sept 2026

Robotics & embodied AI

Robot self-improvement stalls when perception, skill chains and tests mislead

Wang built an agentic robotics system that diagnoses failures, writes skills or installs external models, and tests changes without human-written robot code. The agent successfully discovered some missing capabilities, but its improvements did not accumulate into task success because relational perception, sequential skill testing and the evaluation harness constrained what it could learn.

30 Sept 2026

Robotics & embodied AI

Dual-horizon foresight improves UAV navigation from language instructions

The authors propose ForeFly, a dual-horizon latent world action model that forecasts both proximal and adaptive route-critical future states to aid UAV navigation from natural language. They introduce Foresight-Guided Action Refinement to differentially exploit these forecasts, and show consistent gains over strong baselines on TravelUAV and UAV-ON benchmarks.

30 Sept 2026

Robotics & embodied AI

Symbolic checks make household robot plans safer and cheaper to repair

Bui et al. combine a single LLM planner with a lightweight symbolic verifier and local plan editing, then test it on 75 household tasks in the AI2-THOR simulator. The system matched hierarchical baselines on a saturated standard benchmark, performed better on harder multi-goal tasks, and recovered from injected failures with substantially fewer LLM calls than full replanning.

30 Sept 2026

Robotics & embodied AI

Structured context helps a general vision model control robots better

Liu and colleagues built an inference-time control framework around GPT-6 Astra, organizing demonstrations, actions, and execution feedback into a shared context format. Across 30 simulated tasks, RoboICL scored 50.64 overall, compared with 33.68 for the strongest reported baseline, while small real-robot trials showed substantial gains from demonstrations.

30 Sept 2026

Robotics & embodied AI

Robots Can Revise Visual Plans Instead of Starting Over

Liu et al. introduce Revisable Temporal Planning, a controller that preserves and updates a robot's visual plan rather than always regenerating it after observations diverge from predictions. On two robot-manipulation benchmarks, the method reached task-averaged success rates of 48.6% on RoboMME and 84.8% on RMBench.

30 Sept 2026

Robotics & embodied AI

Federated robot learning preserves local expertise while cutting communication costs

Zhang et al. introduce RoboFL, a federated learning system for combining robot policies trained on heterogeneous, institutionally siloed tasks. Rather than averaging incompatible updates, the system installs local LoRA adapters as specialized experts and learns how to route model inputs among them.

30 Sept 2026

Robotics & embodied AI

Robot-relative 3D features improve pretrained manipulation policies across diverse tasks

Liu et al. introduce Spatial Grafting, which converts features from frozen 3D reconstruction models into robot-relative spatial tokens and feeds them to flow-matching action generators through cross-attention. Across four simulation benchmarks and three physical robot platforms, the authors report consistent gains, with the largest improvements appearing in cluttered and long-horizon manipulation.

30 Sept 2026

Robotics & embodied AI

Predicting future object motion helps robots avoid emerging hazards

Kim et al. present Predictive Semantic Safety, a framework that turns visual predictions of physical events into time-varying occupancy estimates used by a safety controller. In MuJoCo scenarios involving falling objects and contact-driven motion, it kept a simulated Unitree Go1 safe in 99.3% of episodes, versus 43.3% for a baseline limited to current obstacle geometry.

29 Sept 2026

Robotics & embodied AI

Predictive visual models improve action generation for generalist robot manipulation

The authors built a world action model that uses pretrained visual dynamics and geometric knowledge to generate robot actions without rolling out future video during deployment. After pretraining on robot and human demonstration data, they report improvements over prior methods across four simulation benchmarks and multiple real-robot settings.

29 Sept 2026

Robotics & embodied AI

Intermediate targets help frozen visual world models plan longer routes

Liu et al. examine why short-horizon visual planners can fail even when their dynamics predictions and search are accurate. They show that replacing direct goal-image scoring with intermediate targets improves action selection on Cube, PushT, Reacher, and TwoRoom, particularly when successful routes initially move away from the final goal.

28 Sept 2026

Robotics & embodied AI

Rolling predictions make robot replanning 4.5 times faster

Zhou et al. reorganize joint video and action generation so that a robot progressively refines future predictions while executing near-term actions. Rolling-WAM achieved a reported 4.5-fold steady-state replanning speedup over standard joint world action models while maintaining competitive manipulation performance in simulation and on a Unitree G1 humanoid.

28 Sept 2026

Robotics & embodied AI

Coding agents synthesize robot planners that generalize to unseen instances

The authors gave coding agents task descriptions and simulator access, then asked them to develop reusable programs for 28 simulated task-and-motion planning environments. Across 98,000 held-out evaluation episodes, the resulting frozen programs achieved mean success rates of 56% to 95%; on the 16 environments with planner baselines, all three agent configurations exceeded the planners’ 47% mean success.

27 Sept 2026

Robotics & embodied AI

Visual action rehearsal helps language models control robot manipulation

Zhang et al. introduce World Action Agent, a visual workspace in which vision-language models directly propose, rehearse, and correct robot actions using basic control tools. On LIBERO-Pro, the system achieved 75.6% average success, while traces from the harness improved a smaller model’s out-of-domain success from 1.7% to 43.3%.

26 Sept 2026

Robotics & embodied AI

Touch-aware world models improve contact-rich dexterous robot manipulation

Yuan et al. extend a predictive video model with independently encoded fingertip signals, allowing a robot policy to model how contact evolves alongside visual state. On six contact-rich tasks using a 22-degree-of-freedom bimanual platform, DexTacWAM averaged 70.6 points, compared with 38.0 for the strongest baseline.

23 Sept 2026

Robotics & embodied AI

Targeted subtask reinforcement learning improves long-horizon robot manipulation

Su et al. introduce PARTS, a real-world reinforcement-learning framework that leaves a pretrained robot policy frozen while learning corrections for specific failure-prone subtasks. On bimanual YAM and single-arm Franka tasks, complete-task success rose from 32% to 61% and from 50% to 95%, respectively, using tens of minutes of real-world rollouts per task on average.

22 Sept 2026

Robotics & embodied AI

Synthetic demonstrations let robot policies escape sparse-reward failures

The authors introduce SynthDemo-RL, a teacher-student pipeline that generates demonstrations from simulator-privileged state, uses them to fine-tune a vision-language-action policy, and then refines it with reinforcement learning. On LIBERO-PRO, the method achieved nonzero success on all 27 tasks where the starting policy had failed completely, while direct PPO rescued only 10 under matched reinforcement-learning compute.

22 Sept 2026

Robotics & embodied AI

Adaptive action chunking improves robot control by varying horizon based on prediction reliability.

The authors propose GeoAAC, a method for Vision-Language-Action (VLA) policies that adaptively sets the action horizon—the number of future actions predicted at once—based on the geometric properties of the denoising process. By measuring uncertainty through the variation of denoising trajectories across action prefixes, GeoAAC selects a horizon from a single generation without extra training, achieving consistent improvements over fixed horizons in both simulation and real-world tasks.

21 Sept 2026

Robotics & embodied AI

Target masks extend robot policies to ordered, long-horizon manipulation

Huang et al. introduce a harness that connects high-level task planning to short-horizon visuomotor policies through instance-specific image masks. In real-robot experiments, the system completed sequential multi-object tasks more reliably than limited-horizon policies by re-observing the scene and updating the target after each subtask.

21 Sept 2026

Robotics & embodied AI

Handheld demonstrations improve robot policies without repeated robot execution

Han et al. introduce a robot-free, human-in-the-loop method for adapting vision-language-action models after supervised fine-tuning. Across four real-world manipulation tasks, the authors report that policy-guided data collection and progress-based advantage refinement consistently improve performance over supervised fine-tuning.

21 Sept 2026

Robotics & embodied AI

Saliency-driven workspace tokens give robots lightweight task memory

Dashora et al. propose the workspace token, a lightweight latent memory for robotic manipulation trained by using a VLM during training to identify task-salient historical information, then distilling it via set reconstruction. In simulation and hardware, the token serves as a drop-in observation replacement, improving policy performance while eliminating in-the-loop VLM queries.

20 Sept 2026

Robotics & embodied AI

Agent-side memory steers stateless robots through long manipulation tasks

Hu et al. test whether long-horizon robot manipulation requires memory inside the action policy itself. On LIBERO-Mem, their architecture places all interaction history in a multimodal agent and achieves 76.3% average completion with an episodically stateless action model, versus 14.8% for the strongest reported baseline.

13 Sept 2026

Robotics & embodied AI

A semantic harness lets vision-language models control different robots

Chen et al. introduce an interface that lets vision-language models make fine-grained physical decisions using semantic action units, which deterministic interpreters translate into commands for particular robots. The authors report that this approach transfers across tasks, environments, and robot embodiments while outperforming representative agent-based and vision-language-action baselines.

11 Sept 2026

Robotics & embodied AI

Scene-grounded decoding keeps vision-language plans executable and visually supported

Ma and Kordjamshidi introduce CLAMP, a decoding framework that converts visual scene evidence and a symbolic action model into constraints on an embodied planner. Across three benchmarks, the authors report better object grounding and safety, with remaining errors concentrated in perception and incorrectly specified constraints.

10 Sept 2026

Robotics & embodied AI

Adding Depth Helps Video World Models Control Robots More Reliably

Lopetegui Gonzalez, Pacaud, and Schmid adapt a pretrained video diffusion model into a 3D-aware robot policy by training it to predict actions, future images, and depth together. The authors report state-of-the-art performance on RoboCasa and LIBERO-Plus, along with improved results on a real UR5 arm, particularly when environments are randomized.

4 Sept 2026

Robotics & embodied AI

One video world model transfers physical dynamics across robot bodies

Liu and Shorinwa introduce CLAP, an action-conditioned video model designed to learn across different robot morphologies and human demonstrations. The authors report that a curriculum combining unlabeled-video pretraining with robot-action grounding enables zero-shot deployment and performance competitive with specialized, single-embodiment models.

29 Aug 2026