The paper proposes Sensorimotor Alignment (S²), a training-free offline metric that compares a model’s sensory representations with the geometry of physical motion, represented by end-effector trajectories. Across 12 vision-language-action models, S² positively correlates with downstream robot task success. The authors present it as a low-cost way to prioritize models and identify promising vision-language backbones before fine-tuning or costly closed-loop evaluation.
SCOPE is an automated framework for generating diverse, executable task-and-motion-planning datasets for robots. It uses planner execution failures to correct logical inconsistencies in scene graphs, aiming to ensure plans are physically feasible. In experiments, data generated by SCOPE improved a fine-tuned downstream TAMP model’s success rate by an absolute 12.33% over a counterpart trained on DELTA-generated data.
The paper tests whether a Transformer can learn useful robot-action priors from procedurally generated data based on mathematical abstractions of control, without seeing real robot behavior during pretraining. The model achieves non-trivial zero-shot action prediction on unseen physical robot tasks through in-context learning; specialized control priors outperform generic causal priors. A compute-matched model trained on real robot data performs better only on the datasets it saw during training.
Dream Labs announced $90 million in seed funding and launched Dream API, which it says converts a robot camera feed and plain-English instructions into motor commands. Its Dream-1 model is trained on video of people performing everyday tasks and is claimed to transfer across robot hardware without robot-specific data. The company says it currently handles tabletop tasks, including picking and placing objects and folding laundry.
The roundup highlights Workhorse, a system that learns contact-rich whole-body manipulation from human demonstrations without requiring a robot during demonstration. On a real Unitree G1, it sorts boxes using its hands and a kick, catches a thrown box, and topples and climbs a suitcase. The article also features MIT’s SANDO navigation system and a soft underwater robot with three swimming modes.
The video announces the open-source release of Unitree's UnifoLM-WLA-1.0 embodied foundation model, described as coordinating desktop and whole-body mobile manipulation.
