Agility Robotics expects to raise $465 million in its proposed SPAC merger. This capital raise includes funds from Churchill XI's trust account and a private investment in public equity led by Foxconn. The company anticipates this funding will support commercial expansion and growing customer demand for its humanoid robots. Agility plans to close the SPAC transaction on November 4th. The company is working toward commercial deployments and has secured multi-year orders for its Digit 5 humanoid.
RoboBusiness 2026 took place on October 20-21 in Santa Clara, California, attracting over 2,000 attendees interested in robotics developments. Highlights included keynote panels, a Women in Robotics lunch, and an expo floor featuring industry discussions. Sessions addressed topics such as AI in production robotics and operating autonomous robots at scale. The event was the 20th anniversary and the last event under the RoboBusiness brand, now renamed Robotics Summit West. The Robot Operations Group also held RoboOpsCon26 the following day.
Claire interviewed Zlatka Stoeva of DZP Technologies regarding flexible, stretchable, and soft printed electronics for dexterous robots. Stoeva's company specializes in advanced materials and novel electronics, stemming from her background in materials chemistry. DZP Technologies entered robotics through an ARIA-funded project focusing on printed electronics for dexterous robots. The podcast explores the intersection of robotics, AI, and autonomous machines.
Video Friday features robotics videos, including demonstrations of whole-body manipulation. A contact-centric learning framework is presented for versatile humanoid mobility in unstructured environments using motion imitation and reinforcement learning. Luxonis is launching the OAK 4 FFC vision computer for humanoid and drone builders. Other featured projects include robotic hands, industrial navigation systems, and bio-inspired amphibious robots. These selections showcase advancements in robotics, AI, and physical interaction.
Tesla is reportedly planning to open reservations for Optimus in early December, tied to the Humanoids Summit. Modar Alaoui expects most robots will initially be allocated to businesses, with a smaller portion for consumers. Reservations are anticipated to precede commercial deliveries, allowing Tesla to gauge customer interest during development. Details regarding reservation fees and consumer delivery terms remain unclear. Musk's previous guidance suggested external company deliveries would be toward the end of 2027.
Worken is a proposed measure of productive robot operation combining the number of active robots, autonomous working hours, and an output efficiency ratio. It differentiates independent task execution from nonproductive periods like waiting or manual intervention. The formula multiplies these three factors to calculate "Effective Intelligent Work Hours." This measure is not a price for robot labor and requires defining task-specific benchmarks for accurate results. Practical use depends on consistent operational data and further testing to establish reliable definitions.
Robert Scoble's claims about Tesla's Optimus data collection effort have drawn attention regarding the scale and methods. Tesla's own job listings confirm the existence of motion-capture and data collection work across various facilities and shifts. Scoble alleges workers use suits with up to 40 cameras for this data gathering, including recordings of domestic and automotive tasks. While Scoble's details add context, Tesla's postings establish the nature of the required physical and operational work. The text concludes that the actual utility of the data for robot learning remains separate from the collection methods described.
A robotics researcher is surveying where engineering time is spent when moving software from a laptop or simulation to a real robot. The focus is on tasks like integration, bring-up, timing, configuration, and debugging between different components. Participants are encouraged to answer based on the system they spend the most time on, regardless of the specific framework used (e.g., ROS 2, PX4, ArduPilot, or bare-metal). The survey consists of 20 multiple-choice questions and is anonymous. Aggregated results will be shared in the thread approximately two weeks later.
The next Cloud Robotics WG meeting will feature guest speaker Matthew Macdonald-Wallace from Dash0 discussing the next steps after managing logs, advocating for Open Telemetry for metrics, logs, and traces. The goal is to achieve end-to-end visibility from the control platform to the end device. The group is finalizing the Logging and Observability guide, which will be released after final edits and community review. Meeting details and recording links will be provided in subsequent posts. Attendees can find meeting links through the Cloud Robotics Hub or their calendar/Google Group.
Foundry Robotics secured a $100 million Series A funding round led by Marlinspike Partners to build "America's first autonomous factory for assembly." The company builds autonomous factories that combine AI-powered robotic cells with manufacturing expertise. Foundry offers a manufacturing-as-a-service model, allowing customers to go from design to production quickly without retooling. The company focuses on assembling complex electro-mechanical products for the defense and commercial industries. Through its technology, Foundry aims to revitalize US manufacturing by providing scalable and flexible assembly capacity.
Automated cells are bringing inspection closer to the shop floor to improve efficiency. As automation grows, moving Coordinate Measuring Machines (CMMs) next to production allows for rapid part checking. This proximity shortens the feedback loop, enabling immediate adjustments to machining processes. Shared data standards like MTConnect allow robots and inspection equipment to communicate effectively. This integration enables real-time quality control, catching variations early and increasing production confidence.
Bleu Robotics raised €5 million in a pre-seed round to scale its AI models for industrial humanoid robots. The company aims to reduce the humanoid robot deployment process to one hour or less. Bleu Robotics uses an abductive learning model that allows robots to learn new tasks from just ten operator demonstrations, eliminating the need for vast training datasets. This approach enables cost-effective and scalable deployment by circumventing the need for long training cycles. The technology allows operators to teach robots new tasks directly using handheld devices, facilitating easier deployment in manufacturing settings.
Global Ocean Robotics Day convened a symposium exploring the transformation of ocean research through artificial intelligence and autonomous robotics. The event featured talks on topics such as using vehicle trajectories for current mapping, autonomous docking simulations, and underwater robotic swarms. Presentations also covered bio-inspired robots, trustworthy multi-robot systems, and low-cost robotic fleets for ocean deployment. Research highlighted autonomous underwater robot capabilities in perception, navigation, and decision-making using onboard intelligence. The event fostered global collaboration on the intersection of AI and robotics in marine exploration.
Symbotic and Walmart have expanded their commercial agreement, committing approximately $6 billion for further automation. This deal will bring advanced AI-enabled robotics capabilities to 25 US regional distribution centers and support global supply chain growth. Symbotic's solutions, like BreakPack and SymMicro, use AI to optimize the movement of inventory. Walmart aims to use this expanded automation to improve e-commerce inventory management, storage capacity, and product flow. Ultimately, this partnership seeks to enhance customer service by increasing speed and reliability across the supply chain.
Maximo Robotics has completed over 300,000 construction tasks using its US-made, physical-AI robotics systems for utility-scale solar projects. Crew members working with Max 4.0 systems achieved 2.5 times their manual installation rates per person per shift hour. The company operates on a performance-based robotics-as-a-service model where customers pay per completed task. This deployment is helping electricity companies meet energy demand in a labor-constrained market. The platform combines AI models, robotics, and operational intelligence to automate construction tasks across various domains.
Aurora Innovation aims to have over 30,000 driverless trucks operational by 2030 in the United States. The company expects 200 driverless trucks to be operating through Transport as a Service agreements by 2026, with further deliveries scheduled in 2027. Trucks using Aurora's system have completed over 500,000 driverless miles, with customers averaging over 225,000 miles annually. Aurora is increasing hardware production, with Roush currently upfitting trucks at a rate of 20 per week. The autonomous driving technology is currently integrated into International LT Series trucks and is being developed for future integration into PACCAR trucks.
Jungheinrich and Automha formed a strategic partnership to develop a two-way pallet shuttle for high-density storage. This solution combines Automha's shuttle expertise with Jungheinrich's system integration and software strengths. The shuttle is designed to offer an economically viable entry point into warehouse automation by fitting into existing facilities. This technology allows for increased storage density without major structural changes, providing a path toward further automation. The collaboration enables customers to modernize brownfield warehouses gradually, integrating shuttle technology with other automation solutions.
Schaeffler partnered with Wandelbots to develop the Wandelbots NOVA open platform for applying artificial intelligence in industrial and humanoid robotics. This platform aims to enable the use of AI across all Schaeffler plants worldwide for easier development, testing, and transfer of robotics applications. The technology is projected to reduce automation solution life cycle costs by up to 40 percent in manufacturing environments. The partnership will facilitate the scalable use of physical AI and further digitalization across Schaeffler’s global manufacturing network. The joint effort will use simulation technologies to validate applications and create a skill store for systematic knowledge transfer.
AstraQua secured a $1.3 million NASA contract to advance its AstraSense technology for electric and hybrid-electric aircraft. This work focuses on developing Physical AI that incorporates predictive health and operational self-awareness for autonomous systems. AstraSense allows machines to understand their own condition to make safer, real-time decisions, enabling collaborative task management. The technology aims to facilitate adaptation among robot teams, such as in inspection scenarios. This research emphasizes low-power onboard intelligence and health-informed reasoning for autonomous operations.
European autonomous vehicle development is accelerating through expanding robotaxi, automated driving, and autonomous freight programs. Companies like Bolt, Lucid, Mercedes-Benz, and Volkswagen are advancing these technologies, often utilizing Nvidia's computing platforms. Bolt plans to start real-world data gathering for robotaxis in Madrid in 2027 as part of a broader European rollout. This activity spans passenger vehicles and trucking, with a focus on increasing levels of automation. Safety validation, supported by frameworks like Nvidia's Halos SAFE, is becoming a crucial component of this development.
Humanoid is partnering with Nvidia to develop the computing foundation for next-generation AI-powered humanoid robots. They are developing a high-performance robotic edge runtime to allow robots to run sophisticated AI onboard for real-time perception and response. This integration includes Nvidia Jetson Thor, Isaac OS, Holoscan Sensor Bridge, and SDK to process sensor data and coordinate control. This collaboration aims to streamline the architecture and reduce latency for real-world deployment of autonomous robots. Ultimately, the partnership seeks to unlock new levels of autonomy for humanoid robots.
Pantheon Hand is Stella Robot's flagship tendon-driven robotic hand, featuring a lightweight design with electric motors located outside the hand for mass reduction. It has a weight of 0.29 kg, dimensions of 230 × 190 × 50 mm, and a five-finger configuration utilizing single side pull cables with automatic tension adjustment. The hand received an overall score of 7/10, with a higher rating for Degrees of Freedom (4/5) than for Strength (3/5). The design emphasizes dexterity and joint freedom while maintaining a moderate strength assessment. The product is currently in production and is priced at USD 3,500.
Mantis Standart, nicknamed Xiaobai, is a modular wheeled humanoid robot developed by Blue Worm in Xi'an, China. It is designed for research, data collection, and light industrial work, standing 140 cm tall and weighing 50 kg. The robot features two articulated arms, an adjustable waist, and supports up to 22 degrees of freedom for tasks like sorting and assembling. It is supported by ROS 2, MoveIt 2, and VR teleoperation, offering an accessible entry point into wheeled humanoid development. The platform is listed as in production with a starting price of US$1,400.
Lumen H1 is a 169 cm tall, 69 kg humanoid robot developed by Guanggu Dongzhi in China. It possesses over 40 degrees of freedom and hands with five fingers, capable of obstacle avoidance and sorting goods. The robot has a Humanoid.Guide Skill Score of 2/10, with navigation and manipulation rated 1/5. Safety features include LiDAR, RGB depth cameras, and PLd level protection with a braking distance under 1.7 meters. The robot is available in silver and yellow and is marketed for service environments at a price of USD 40,000.
Laborer C1 is a compact, bipedal humanoid robot from Glroad, measuring 138 cm and weighing 38 kg, equipped with electric joint actuators. It demonstrates capabilities like obstacle avoidance, 100m running, jumping, landing, and sorting goods. Navigation is rated better than manipulation, suggesting a stronger locomotion profile. The robot is listed as in production with a price of USD 70,000 and is positioned for public interaction. Its utility lies in demonstrating movement skills within human environments rather than complex industrial manipulation.
TIAGo Pro is a humanoid mobile manipulator from PAL Robotics designed for research and applications in various fields. It features a humanlike upper body, two arms, a touchscreen face, and an omnidirectional wheeled base. The robot stands 155 cm tall, weighs 80 kg, and offers capabilities for obstacle avoidance, sorting, and product assembly. It incorporates torque sensing for compliant interaction and runs on an Ubuntu LTS operating system with ROS 2. The robot is currently in production with a listed price of USD 90,000.
Lumen S1 is a compact 130 cm, 35 kg humanoid robot developed by Guanggu Dongzhi in China. It features 31 body degrees of freedom with electric joints and five-fingered hands, capable of obstacle avoidance and sorting goods. The robot is equipped with NVIDIA Jetson Orin NX computing and various sensing hardware including LiDAR. Its performance ratings place it in an early stage category, with preliminary navigation and manipulation skills. The robot is listed for production at a price of USD 20,000 and targets markets like education and research.
AgiBot generated nearly 2 billion yuan in revenue in the first half of 2026, reflecting rising humanoid shipments. The company shipped 9,700 humanoid robots in that period and claimed a 43.1 percent share of global shipments. AgiBot emphasized that its figures demonstrate the ability to manufacture at scale and deploy robots effectively, emphasizing the need for supply chains and maintenance. Unitree Robotics also reported strong financial results, with revenue growing significantly in the first nine months of 2026. Revenue growth is attributed to increased manufacturing capacity, expanded domestic and overseas applications, and the integration of long-term service solutions alongside hardware sales.
Most existing robotics data consists only of successful robot behaviors. Research areas like world models and reward model learning require examples of failure, such as unsuccessful manipulation trajectories. Claas Voelcker, Arpit Bahety, and Paul Zhou are presenting their work on "oopsie data," which involves collecting these valuable failures. Their goal is to create the first large, community-driven open-source dataset of failure examples. This effort aims to provide necessary data for advancing robotics research.