No Priors Ep. 48 | With Covariant CEO Peter Chen
Covariant co-founder and CEO Peter Chen joins Sarah on No Priors to discuss building robotics foundation models, deploying robots in warehouses, and the data and safety challenges involved.
Yun Jia Pan, founder of Wuji Technology, discusses direct-drive dexterous hands, data collection, and commercialization with Peter Chen of Alphaist Partners on Wuji’s Talk podcast.
Pan says he revisited dexterous hands after an unsuccessful home prototype in 2018 and saw an opening when Wuji shifted focus in late 2023. Drawing on the company’s motor expertise, he reasoned that motors small enough for finger joints could exceed human torque and speed. He argues tendon-driven designs add complexity through paired motors or springs, require forearm integration, and can face wear and simulation challenges.
Wuji benchmarks its hand against human anatomy across dimensions including degrees of freedom, size, weight, force, speed, sensing, and reliability. Pan describes reliability as difficult to measure with a single test because different grips load different joints. He says quality needs company-wide ownership from design through manufacturing and customer feedback, with reliability expertise involved early.
Pan says system identification can make simulated and physical hand response closely match. He favors a lightweight glove worn by people for collecting manipulation data, arguing that exoskeletons such as DexYumi and DexOp are better suited to supplementary training data than large-scale collection. Wuji’s approach uses motor-current feedback for force control, and Pan expects a unified large model trained with enough data to handle manipulation tasks.
Pan identifies data acquisition as a key obstacle to deployment and expects dexterous hands to begin scaling in industry around 2027. Wuji is focusing on research, education, and algorithm companies in 2026, then industrial users, with consumer markets later. He says the company plans to open-source robot-arm adaptations and provide an SDK, while prioritizing hardware and data-collection tools over training its own advanced models.
Pan describes direct-drive hands as a systems-engineering challenge where every component must work well, and says execution speed is a startup’s key advantage. He says Wuji has operated with limited funding and wants to invest in critical in-house processes, data equipment, and talent. He sees integration across reliable hardware, data, and alignment between human and machine behavior as a route toward intelligent robots.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.
Covariant co-founder and CEO Peter Chen joins Sarah on No Priors to discuss building robotics foundation models, deploying robots in warehouses, and the data and safety challenges involved.
PRO ROBOTS discusses Wuji Technology’s dexterous robot hands, data collection glove, and their potential role in embodied AI at ICRA 2026 in Vienna.
Yunja Pan, founder of Wuji Technology, discusses direct-drive dexterous hands, simulation, data collection, and company strategy with Peter Chen of AlphaFund Partners.