Gasgoo Munich-Singapore-based AI robotics company Sharpa has announced its Wave dexterous hand served as core verification hardware. Mounted on Apptronik's Apollo 2 robot, it completed multi-finger dexterous tasks such as tying trash bags, unscrewing light bulbs, and sealing resealable bags.
Image Credit: Sharpa
This demonstration project aims to highlight the capabilities of Google DeepMind's new model. In late July, Google DeepMind officially released the new generation robot foundation model, Gemini Robotics 2. The core breakthrough lies in further unifying vision, language, and motion control, enabling humanoid robots to coordinate whole-body movements.
Google notes that unlike previous models focused mainly on upper-body movement, Gemini Robotics 2 achieves end-to-end control of the entire humanoid body—from feet to fingertips—and supports multi-robot collaboration.
In a pre-recorded demonstration, DeepMind's new model guided Apptronik's humanoid robot, Apollo, across a room to pick up a watering can and place it on a shelf—all while navigating around obstacles.
Meanwhile, Google DeepMind simultaneously released two additional robot AI models: Gemini Robotics ER 2 and On-Device 2. Both can work together or run independently, supporting the planning of multi-step tasks and the coordination of multiple robots.
Founded in 2024 by Hesai Technology's three co-founders—Li Yifan, Xiang Shaoqing, and Sun Kai—Sharpa is a rising player in the embodied intelligence sector that vaulted into unicorn status immediately upon launch. The company's global headquarters is in Singapore, with manufacturing and R&D centers in Shanghai, China, and commercial operations headquartered in Mountain View, California.
Image Credit: Sharpa
As Sharpa's inaugural product, the Sharpa Wave high-performance tactile AI dexterous hand features 22 active degrees of freedom per hand, with independent control for all five fingers. Integrated with more than 1,000 tactile sensing units at the fingertips, it provides the hardware foundation for fine grasping, force perception, and real-time adjustment—making the mapping of human motion to robot movement far more natural.
Currently, the Sharpa Wave has become standard equipment for top-tier global robotics research institutions. Beyond Google DeepMind, cutting-edge labs and elite universities such as NVIDIA GEAR Lab, MIT, Stanford, UC Berkeley, Tsinghua University, and Peking University have all adopted the solution. It has supported frontier embodied intelligence research, including EgoScale and Gemini Robotics 2, establishing itself as a critical hardware interface connecting AI models with the physical world.
Beyond the lab, Sharpa is pushing forward with commercial deployment.
According to Sharpa, in August the company will partner with DQ in Shanghai to create the world's first "zero-modification, fully autonomous" robotic dining space. There, Sharpa robots will serve as resident staff, using DQ's existing ingredients, equipment, and workflows to autonomously complete the production and delivery of Blizzard ice cream treats.
