Tags :WITA-Omni

AGIBOT releases WITA-Omni embodied multimodal robot interaction model

Chinese robot maker AGIBOT releases WITA-Omni, an end-to-end embodied omni-modal interaction model trained on robot proprioceptive data. The model supports audio and video inputs and outputs, adds tactile and proprioceptive state inputs, and outputs robot actions and facial expressions, according to the company. Built on a streaming full-duplex architecture, it can listen, think, respond, move and make facial expressions simultaneously, with response latency as fast as about 400 milliseconds and averaging about 600 milliseconds — nearly three times faster than the industry average, AGIBOT said. Its omni-modal rejection capability eliminates the need for wake words by combining conversational orientation, video, audio, voiceprint, lip movement, posture, semantics and context to decide whether a current utterance needs a response, reducing false triggers...

2026-11-11Briefs

Briefs

AGIBOT releases LIFE lifelong embodied intelligence framework
Chinese robotics company AGIBOT release LIFE, a lifelong embodied intelligence framework that combines continuous perception, long-term memory and skill reuse. The framework enables robots to gather environmental information during and between tasks and to use past observations to support subsequent decisions. LIFE has three core components: RoboBeats, a lightweight decision model for condition tracking and event discovery; Plastic Memory, a plasticity memory module that stores raw footage and structured records and supports retrieval and information supplementation; and EDM, a capability reuse mechanism that turns methods developed during problem-solving into validated tools and skills for later tasks.
8 hours ago
AGIBOT releases WITA-Omni embodied multimodal robot interaction model
Chinese robot maker AGIBOT releases WITA-Omni, an end-to-end embodied omni-modal interaction model trained on robot proprioceptive data. The model supports audio and video inputs and outputs, adds tactile and proprioceptive state inputs, and outputs robot actions and facial expressions, according to the company. Built on a streaming full-duplex architecture, it can listen, think, respond, move and make facial expressions simultaneously, with response latency as fast as about 400 milliseconds and averaging about 600 milliseconds — nearly three times faster than the industry average, AGIBOT said. Its omni-modal rejection capability eliminates the need for wake words by combining conversational orientation, video, audio, voiceprint, lip movement, posture, semantics and context to decide whether a current utterance needs a response, reducing false triggers during multi-person conversations or environmental broadcasts.
11 hours ago
D-Robotics, ROKAE partner on edge compute for dual-arm and humanoid robots
D-Robotics and ROKAE (3752.HK) announced a deepening partnership under which D-Robotics' Sunrise S600 edge computing platform will provide integrated compute support for intelligent decision-making, multimodal perception and real-time control in ROKAE's AR5 force-controlled dual-arm integrated form and Zhuxing H-series humanoid robots. ROKAE's Xuanhe iNex open motion control system is adapted to the Sunrise S600, allowing the AI brain and whole-body motion control to run on the same machine without a separate motion control box.
11 hours ago
Anthropic releases flagship Fable 5.2, touts robot-control performance
Anthropic released its new flagship model, Fable 5.2. In the RoboDojo dual-arm simulated robot benchmark, Fable 5.2 achieved an average success rate of 64.9%, controlling a physical robotic arm to complete three tasks — drawing patterns, unscrewing a bottle cap and solving a Rubik's cube — without dedicated training, according to test data published by the company. Anthropic said Fable 5.2's robot-control performance is "better than most AI models designed specifically for robot control."
12 hours ago
UBTech wins 88.9 mln yuan Shaoxing robot education project
UBTech, known as China's "first listed humanoid robot company," has won a bid for a Shaoxing embodied intelligence industry-education integration center project, according to a disclosure by Zhejiang Enterprise Procurement Information Service Network. The winning bid was 88.892378 million yuan ($12.3 million). The procurement covers primary and intermediate education humanoid robots, building-block robot teaching kits, modular mobile chassis robot kits and robot controllers, among other items.
1 days ago
Ant Lingbo releases tactile foundation model LingBot-Touch
Ant Lingbo Technology released LingBot-Touch, a tactile foundation model pretrained on nearly 5,000 hours of real tactile data. It converts magnetic, vision-based and piezoresistive signals into a unified representation for dexterous hands and action models, adding touch sensing for fine manipulation. The model and code are open-sourced on Hugging Face, GitHub and ModelScope.
1 days ago
Anthropic withdraws pricing details from IPO filing sent to bankers
Anthropic sent bankers working on its initial public offering a filing that included pricing information, then withdrew those details, according to the document and a subsequent email. The filing initially said the company planned to offer about 867 million shares at $70 to $80 apiece. Anthropic later said in an email that the pricing information was inaccurate and that it would update the filing later.
1 days ago
ACE ROBOTICS deploys dual-brain architecture in hotel laundry
ACE ROBOTICS said it has deployed its dual-brain collaborative architecture in a hotel laundry scenario for the first time, with embodied recursive self-improvement (RSI) enabling continuous skill evolution. The deployment brings together world models, general-purpose models and embodied RSI.
1 days ago