Technology areas
Robot models and learning
Robot VLA models, general action and world models, imitation and reinforcement learning, skill and motion policy learning, online adaptation, and transfer across tasks and robot bodies. The contribution is a model, policy, or learning method. Dataset releases belong under Data and simulation; inference deployment belongs under Software and compute systems.
206 selected · 116 in the last 30 days
Figure AI ·
Figure AI shows two Helix-02 humanoid robots collaborating to tidy a bedroom
Figure AI demonstrated two humanoid robots equipped with Helix-02 collaborating to tidy a bedroom within two minutes, completing tasks such as opening a door, hanging clothes, putting away headphones, closing a book, taking out trash, pushing a chair under a desk, and making a bed together. The two robots run the same learned vision-language-action (VLA) policy, with no shared planner, message passing, or central coordinator; each infers the other's intent from its actions via its own cameras. The company says that to its knowledge this is the first demonstration of multi-humanoid collaborative mobile manipulation completed by a single learned neural network directly from pixels to actions.
LimX Dynamics ·
LimX Dynamics Open-Sources FluxVLA Engine, a Standardized VLA Engineering Foundation
LimX Dynamics has officially open-sourced FluxVLA Engine, a standardized VLA engineering foundation for embodied intelligence that covers the full pipeline from data and training to simulation evaluation and real-robot deployment.
Figure AI ·
Figure 03 production ramps up: BotQ capacity rises 24-fold in 120 days to 1 unit per hour
Figure announced that its BotQ factory has delivered more than 350 third-generation humanoid robots, Figure 03, with takt time improving from 1 unit per day to 1 unit per hour, a 24-fold throughput increase in 120 days. The line is driven by self-developed manufacturing execution software and covers more than 150 networked workstations. First-pass yield for completed units exceeds 80%, the battery line's first-pass yield is 99.3%, more than 9,000 actuators have been produced, and each robot undergoes more than 80 functional verification tests before leaving the factory.
New Scientist Robots ·
Sony AI table tennis robot Ace defeats professional players; research published in Nature
Sony AI's autonomous table tennis robot Ace competed under the rules of Japan's professional table tennis league, defeating multiple elite and professional players; the research was published in Nature (DOI: 10.1038/s41586-026-10338-5).
PsiBot/灵初智能 ·
Lingchu Intelligence releases Psi-R2 and Psi-W0, open-sources 1,000-hour full-modality human hand manipulation dataset
On April 10, Lingchu Intelligence released Psi-R2, pretrained on human data, and the Action-Conditioned World Model Psi-W0, and open-sourced the first 1,000 hours of a full-modality human hand manipulation dataset, with total data reserves reaching 100,000 hours.
Physical Intelligence ·
Physical Intelligence releases steerable general-purpose robot model π0.7, demonstrating compositional generalization
Physical Intelligence has released the general-purpose robot model π0.7, which it says can perform a variety of dexterous manipulations with speed and robustness comparable to fine-tuned specialized models, and shows preliminary signs of compositional generalization on tasks not seen in its training data.
Skild AI ·
Skild AI Acquires Zebra Technologies' Robotics Division to Advance Warehouse Deployments
Skild AI announced the acquisition of Zebra Technologies' robotics division (formerly Fetch Robotics), and plans to deploy Skild Brain in warehouse settings to automate end-to-end fulfillment processes. Zebra's Symmetry Fulfillment scheduling platform has been validated in logistics environments, where it can coordinate tasks among robots and frontline workers; the company says the acquisition will accelerate its data flywheel and expand into more verticals and robot form factors.
PsiBot/灵初智能 ·
Lingchu Intelligence Releases Psi-R2 and Psi-W0 Dual Models, Pretrained on 100,000 Hours of Human Data
Lingchu Intelligence released two models, Psi-R2 and Psi-W0, which it calls the industry's first embodied intelligence models pretrained on human data at the 100,000-hour scale.
Zerith Robotics/零次方 ·
Zero Power Robot completes over 100 million yuan funding round led by Runze Group
Zero Power Robot announced the completion of a new funding round of over 100 million yuan, led by Runze Group, with Ningbo Dongli, Jieli Angel, and Pinghu Zexin participating. The company said its order count has exceeded 100 million yuan and revenue is in the tens of millions, with stable monthly mass production of 100 units, and validation completed in more than 20 landmark scenarios nationwide; the article also introduced its technological progress including the 4D augmented model ZERITH-SDA and the brain-inspired dual-stream asynchronous execution architecture ZERITH-V2.
Agile Robots/思灵机器人 ·
Agile Robots and Google DeepMind announce strategic partnership to integrate Gemini Robotics models into industrial robots
Agile Robots SE and Google DeepMind announced a strategic research partnership to combine the Gemini Robotics foundation models with Agile Robots' industrial robot platform and develop adaptable, reasoning-capable robots. The partnership initially focuses on industrial manufacturing scenarios where reliability and scale are critical, forming a data flywheel through deployment, data collection, model training, and iteration; Agile Robots says it has installed more than 20,000 robot solutions worldwide.
Berkeley AUTOLAB ·
CaP-X: Benchmarking and Improvement Framework for Robot Manipulation Coding Agents Released
Teams including Berkeley AUTOLAB propose CaP-X, an open framework for systematically studying Code-as-Policy agents in robot manipulation, centered on the interactive environment CaP-Gym and CaP-Bench, which evaluates 12 frontier models.
Dobot/越疆机器人 ·
Dobot begins mass production delivery of third batch of full-size industrial humanoid robot ATOM
Dobot Robotics has officially begun mass production delivery of the third batch of its full-size industrial humanoid robot ATOM, deploying it to frontline industrial applications worldwide. ATOM runs on the self-developed DOBOT-VLA model, which combines vision, language, and action for autonomous decision-making; its joint modules have undergone nearly 10,000 cycle tests, and it supports coordinated scheduling with wheeled and multi-legged robots on a unified platform.
Physical Intelligence ·
Physical Intelligence proposes RLT: efficient online reinforcement learning for fine manipulation
Physical Intelligence proposes a new method called RL tokens (RLT), in which a vision-language-action model (VLA) outputs a special token to interface with a lightweight online reinforcement learning policy, improving the speed and success rate of fine manipulation phases with only a small amount of real data.
Skild AI ·
Skild AI partners with ABB Robotics, Universal Robots, and NVIDIA to deploy Skild Brain
Skild AI announced partnerships with ABB Robotics and Teradyne-owned Universal Robots and MiR to integrate its general-purpose model Skild Brain into industrial robots, eliminating task-by-task programming and using post-training fine-tuning with a small amount of task data.
RoboForce ·
RoboForce closes $52 million funding round, bringing total funding to $67 million
RoboForce announced an oversubscribed $52 million funding round, bringing its cumulative funding to $67 million. The round was led by YZi Labs, with participation from Jerry Yang. The new capital will be used to accelerate next-generation robot foundation models, scale general-purpose Physical AI robots, and advance manufacturing readiness for commercial deployment.
Physical Intelligence ·
Physical Intelligence proposes MEM, enabling π0.6's VLA to remember long-horizon tasks of up to 15 minutes
Physical Intelligence proposed a method called Multi-scale Embodied Memory (MEM) that provides vision-language-action models (VLAs) with both short-term and long-term memory, enabling the model to complete tasks lasting up to 15 minutes, such as cleaning an entire kitchen or making a grilled cheese sandwich from scratch.
普华资本 ·
Boundless Dynamics completes over 200 million yuan Angel+ round, third round nearing completion
General-purpose embodied intelligence robotics company Boundless Dynamics announced the completion of an over 200 million yuan Angel+ funding round, jointly invested by Puhua Capital, Shenzhen Capital Group, Innovation Works, Junshan Capital, and Yari Capital, with existing shareholders Sequoia China, Linear Capital, and Hillhouse Venture Capital increasing their investments. The third round of financing is nearing completion, and cumulative funding will reach 800 million yuan.
Physical Intelligence ·
Physical Intelligence proposes a physical intelligence layer and publishes π 0.6 deployment data from Weave and Ultra
In a blog post, Physical Intelligence argues that robots need a reusable physical intelligence layer, and it published real-world deployment data from partners Weave and Ultra using π 0.6.
线性资本 ·
Interview with Tashi Intelligence's Chen Yilun: No VLA, no simulation—an embodied AI company's contrarian bets
Linear Capital reposted LatePost's interview with Tashi Intelligence founder Chen Yilun, who laid out a technical path that differs from the mainstream: he rejects the VLA architecture that grows an action head on top of a VLM, and instead develops its own AWE (AI World Engine), aiming to represent physical quantities such as time, space, and force; it has also abandoned teleoperation data collection in favor of a self-developed collection device combining a wearable glove and a first-person-view camera.
Figure AI ·
Figure AI Releases Helix 02: A Full-Body Autonomous Control Model for Humanoid Robots
Figure AI has released Helix 02, extending control from the upper body to the entire humanoid robot and enabling continuous, integrated autonomous tasks combining locomotion and manipulation. It adds a new System 0 whole-body controller, trained on over 1,000 hours of human motion data and simulation-based reinforcement learning, which outputs joint commands at 1 kHz via a 10M-parameter network, replacing roughly 109,504 lines of hand-written C++ code; System 1 connects head-mounted vision, palm cameras, fingertip touch sensing, and proprioception, outputting whole-body joint targets at 200 Hz.