RoboSignal Daily · 2026-11-05 · Thursday
ROBOSIGNAL
Meta reportedly collaborates with Chinese robotics firm Fourier on humanoid robot 'Max'; Fourier responds pending official announcement
On November 1, foreign media cited unnamed sources claiming Meta is collaborating with Chinese robotics company Fourier to develop a humanoid robot prototype named 'Max', currently in its second generation.
Robot Models
WayTrace: A System Using Off-the-Shelf VLM, Onboard Trace Model, and RL Policy for Real-World Navigation
The ETH Robotic Systems Lab introduces WayTrace, a robotics system that converts web-map directions into real-world navigation tasks. The system utilizes an off-the-shelf Vision-Language Model (VLM), an onboard trace model, and a Reinforcement Learning policy to guide a robot through complex environments in Zurich, covering a total distance of 2.35 km. This demonstration includes navigating crosswalks, stairs, crowded streets, and construction sites.
ROS Discourse — Community updatesplynth: connect every robot to every building
Peter Pasha introduces plynth, a new open-source API and ROS 2 toolkit designed to simplify the integration of robots with building infrastructure such as elevators and doors. The tool aims to standardize interactions between robots and various building systems, reducing the need for custom integrations at each new site. It includes Python clients, ROS 2 nodes, and adapters for Open-RMF, allowing robots to request passage through doors and ride elevators under rules set by building owners. A sandbox environment is available for testing without sign-up, and the project seeks feedback from robot teams before deploying in real buildings.
ROS Discourse — Community updatesBrowser-based BehaviorTree.CPP editor with Nav2 import/export
The author maintains behaviortrees.com, a free, open-source visual behavior tree editor running in the browser. It imports and exports BehaviorTree.CPP XML (v4 for Jazzy and Groot2, v3.8 for Humble), including Nav2's navigate-to-pose trees. The tool includes a CLI for CI pipelines to validate tree structures (e.g., catching dangling children or cycles) and convert between BT.CPP versions and behavior3 JSON. An MCP server allows AI assistants to draft, validate, and share trees via links. It is designed to complement Groot2 by requiring no installation, making it suitable for quick sketching, reviewing, or sharing. The project seeks feedback from Nav2 users regarding missing features and round-trip reliability.
Research
arXiv Robotics — research abstractsReflexArc: Bottom-Up Integration of Instantaneous Tactile Reflexes into Whole-Body Humanoid Control
This paper introduces ARC (Ascending Reflex Control), a framework that integrates instantaneous tactile reflexes with learned whole-body control for humanoid robots. The system detects hazardous contacts via full-body tactile sensing, solves withdrawal poses within 10 ms, and combines reflex-in-the-loop policy learning to maintain balance and complete recovery tasks. Evaluated on a Unitree G1, ARC reduces the first force peak by 34.1% compared to a delayed reflex baseline and lowers contact impulse from 0.2632 N s to 0.1225 N s. The authors also demonstrate protected motion tracking, teleoperation, and vision-free obstacle traversal guided by foot tactile sensing.
MM-ABC Enables Mobile Manipulation by Solving Dynamic Perception and Coordination Challenges
Researchers from HKUST (Guangzhou) and Tsinghua University propose MM-ABC, a foundation model for general mobile manipulation. The system addresses dynamic spatial perception and arm-base coordination through clear action prediction, multi-layer visual feature injection, and a future expert module used during training. Trained on over 5000 hours of data including the self-collected MM-30 dataset, MM-ABC achieves a 44.71% success rate on EBench and 61.2% on RoboCasa365, outperforming baselines like π0.5 in both fixed-base and mobile manipulation tasks.
RobohubRobots in society, business and culture: October 2026
This article compiles robotics news from October 2026, covering a biohybrid robot that moves upon electrical stimulation, a study showing people perceive robots approaching from behind as faster, an AI system enabling monocular drone racing at extreme speeds, lab-grown muscle tissue powering bioactuators with increased range of motion, a quadruped robot solving a Rubik's Cube autonomously, RobCo reaching a USD1 billion valuation ahead of its Alfie robot launch in March 2027, an expanded partnership between GXO Logistics and Agility Robotics for Digit humanoid deployments, the opening of TUM RoboGym with 48 training cells for embodied AI, a partnership between Teradyne Robotics and Sereact for industrial AI solutions, and an art exhibition using wobbling humanoid robots to explore social interaction.
Tech Xplore — RoboticsUnderwater exoskeletons could help divers use less air and move more efficiently
Researchers at the Shenzhen Institute of Advanced Technology (SIAT) developed PEAKED, a soft underwater hip exosuit that assists leg movements to reduce physical effort and air consumption for scuba divers. Published in Science Robotics, the study reports that the device reduced respiratory minute volume by up to 29.77% during flutter kicking in human experiments. Unlike rigid structures, PEAKED uses a compliant design to support natural lower-limb movements while allowing divers to retain full control and situational awareness. The team highlights future directions including personalized assistance and posture stabilization to transform these devices into intelligent wearable systems for demanding underwater tasks.
Data & Training
arXiv Robotics — research abstractsSomaAlign: Reliability-Aware Multimodal Representation Learning for Soft Robotic Grasping
This paper introduces Soma, a multimodal dataset for soft robotic grasping with 100 objects and 74,914 samples across RGB, depth, tactile, IMU, and text modalities. The authors also release Soma-Success, an independent grasp-outcome dataset of 1,000 trials used to evaluate frozen-encoder transfer for post-hoc classification and early prediction. SomaAlign is a representation learning framework built on Soma that uses masked reconstruction, reliability-aware contrastive alignment, and conflict-aware semantic constraints. On 30 cross-modal retrieval tasks, SomaAlign achieves an average Recall@10 of 90.66%. When applied to Soma-Success with lightweight probes, frozen SomaAlign encoders reach 94.59% post-hoc classification accuracy and 84.07% early prediction accuracy. The work provides a foundation for robust multimodal perception in soft robotic manipulation.
Robots & Hardware
Cainiao Climbing Robot Showcased at Shanghai; Multiple Benchmark Projects Delivered Globally
On November 3, Cainiao's self-developed climbing robot system was unveiled at CeMAT ASIA 2026 in Shanghai, highlighting its four-sided retrieval and dual-depth storage capabilities along with new robot scheduling and warehouse planning simulation systems.
Deployments
Six Zhiyuan Genie G2 Robots Begin 24-Hour Work Trial at Shanghai Convenience Store
On November 2, six Zhiyuan Genie G2 humanoid robots started a 24-hour operational trial at a convenience store in Xuhui District, Shanghai. Assigned roles included greeter, barista, stocker, cashier, and cleaner. The robots faced real-world challenges such as navigating obstacles, handling delicate items like rainbow candy without crushing them, and distinguishing stains from product patterns. Their performance metrics including task success rate, response speed, and customer feedback were monitored for a 'probation-to-permanent' assessment.
Industry
Meta reportedly collaborates with Chinese robotics firm Fourier on humanoid robot 'Max'; Fourier responds pending official announcement
On November 1, foreign media cited unnamed sources claiming Meta is collaborating with Chinese robotics company Fourier to develop a humanoid robot prototype named 'Max', currently in its second generation.
Xiaomi Wins Dual Awards at 2026 World Internet Conference Wuzhen Summit
On November 3, 2026, Xiaomi Group received the 'Outstanding Contribution Award' and its general-purpose robot operation base model technology was awarded the 'Leading Technology Award' at the 2026 World Internet Conference Wuzhen Summit. This marks Xiaomi's first time winning both awards simultaneously.
Briefs
- RoboColosseum: Evaluating Physical Intelligence at Scale In-The-WildarXiv Robotics — research abstracts
- Behavior Graphs: Unsupervised Graph Structures for Interpreting Robot Deployment DataarXiv Robotics — research abstracts
- Design and Experimental Validation of the Separation and Deployment System for the CHERI Lunar Micro-RoversarXiv Robotics — research abstracts
- Action-Consistent Decentralized Belief Space Planning with Inconsistent Beliefs and Limited Data Sharing: Framework and Simplification Algorithms with Formal GuaranteesarXiv Robotics — research abstracts
- h-NESDF: An Infinitely Differentiable Signed Distance Between Two Sets with Robotic ApplicationsarXiv Robotics — research abstracts
- CoReMatch: Recurrent Continuous Coordinate Refinement for Real-Time Robotic Feature MatchingarXiv Robotics — research abstracts
- MESSPa: A Multi-modal Embodiment Stimulation Shoulder Pad for Preserving the Sense of Agency in a Detachable Robotic ArmarXiv Robotics — research abstracts
- Towards end-to-end co-design for one-shot printable task-specific soft robotsarXiv Robotics — research abstracts
- Swarm Choreography Made Simple: Superposed Guiding Vector Fields for Rigid Formation Path FollowingarXiv Robotics — research abstracts
- HuG: Learning Humanoid Whole-Body Affordance-Aware Grasping with Active PerceptionarXiv Robotics — research abstracts