Ep#40: Daxo Robotics with Tom Zhang
Tom Zhang, founder and CEO of Daxo Robotics, discusses the company’s tendon-driven robotic hand and its design, capabilities, and possible applications with the RoboPapers hosts.
Tom Zhang of Daxo Robotics discusses robotic hand design, the company’s technology and its path from student research and agricultural products to a dexterous hand at a GRASP Lab talk at Penn.
Zhang argues that many useful tasks depend on hands, and that dexterity remains difficult even as AI advances. He compares three designs: linkage hands that are affordable and robust but limited, direct-drive hands with many motors but high cost and failure risk, and tendon-driven hands that offer dexterity but face expense and tendon wear.
Zhang questions whether human hand movement is adequately described by counting degrees of freedom, pointing to subtle joint motion and the way fingers adapt when carrying weight. Daxo’s first prototype used 108 motors, including 20 per finger, to drive a tendon-based continuum design with force sensing. He reports a roughly $1,200 prototype, a 5-kilogram payload per finger, and a 100,000-cycle stress test with no visible tendon wear.
Zhang says force measurements from the tendons can provide useful information about finger contact, though they do not replace high-resolution tactile sensors. The team used MuJoCo to explore feasibility but has not deployed its policies through sim-to-real transfer; it is also collecting real-world demonstration data for tasks such as drawing circles. A planned 120-motor version adds a wrist, fits a forearm form, and is intended for launch later in 2026.
Zhang traces the company’s direction to PhD work on swarms, chaotic systems and soft robotics, along with market discovery that led to apple picking. Daxo developed an apple stem clipper and sold nearly 600 units across New Zealand, Chile and U.S. apple-growing states. In March 2025, despite having more than 40 customers, Zhang pivoted the company toward the robotic hand, citing the need for greater dexterity and faster iteration.
Zhang says persistence helped him start the company, while he wishes he had begun sooner. He describes manufacturing more than 100 tendons as a current bottleneck and says the team is working on ways to assemble them faster; the hand is designed for easy motor replacement and a replaceable molded hand section. Potential applications raised include wire harness assembly and packing fragile berries, while Zhang says human-hand retargeting remains a challenge for data collection.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.
Tom Zhang, founder and CEO of Daxo Robotics, discusses the company’s tendon-driven robotic hand and its design, capabilities, and possible applications with the RoboPapers hosts.