TARS has capped a landmark look at ICRA 2026, IEEE’s worldwide robotics convention, with the worldwide debut of its DexHand platform, drawing important consideration amongst industrialists and lecturers.
Dr Ding, TARS’ chief scientist and co-founder, delivered the keynote tackle within the plenary session.
TARS’ DexHand demonstration showcased all 26 English alphabet sign-language gestures and invited to have interaction in real-time mirror-control interplay, thus providing dwell proof of the system’s biomimetic constancy and low-latency responsiveness.
On the coronary heart of DexHand is a 21-DoF structure modelled 1:1 on human metacarpal and phalangeal topology. Not like typical parallel-joint designs that introduce kinematic distortion throughout advanced actions, DexHand replicates the spatial convergence of the thumb’s CMC and MCP joints, eliminating the movement blind spots.
Its self-developed joints combine high-precision reducers, decreasing backlash to an especially small vary and delivering silky-smooth micro-manipulation.
This biomimetic construction additionally solves certainly one of embodied AI’s most urgent bottlenecks: the hole between simulation and actuality. TARS’ SenseHub captures information from actual human movement and map it, thus bettering information utilization with none loss.
DexHand’s fingertips combine ultra-high-resolution miniature digicam modules able to capturing microscopic textures as wonderful as 0.05mm at over 240Hz.
Its AWE 3.0 embodied basis mannequin, allows the robotic to grasp bodily properties comparable to hardness, roughness, and slip threat, and to foretell prevalence fairly than merely reacting after the actual fact.
On the manufacturing facet, DexHand’s inflexible quasi-direct-drive design, utilizing simply three motor varieties and reducer varieties, is purpose-built for automated meeting traces.
“IEEE’s ICRA 2026 was the perfect stage to showcase TARS’ embodied AI options in observe,” mentioned TARS’ chief scientist and co-founder Dr Ding. “TARS’ DexHand is the optimized interface between human intelligence and robotic motion.”
