Yogi
Yogi is an emotionally expressive social humanoid robot by Cartwheel Robotics (Reno, NV), designed for home and healthcare companionship. Features medical-grade silicone skin, 50+ DOF, and a Motion Language Model (MLM) for personality-driven gestures. Designed by ex-Boston Dynamics and Disney Imagineering engineer Scott LaValley.
Primary use case: Natural human-robot interaction with expressive and responsive movement.
Specifications
Physical
Locomotion
unknown
Manipulation
Hand Type
unknown
Max Lift/Arm
15 kg
Power & Autonomy
Autonomy Level
unknown
Commercial
Availability
unknown
Primary Use Case
Natural human-robot interaction with expressive and responsive movement.
Related news
Vesoma Emerges From Stealth With a Humanoid Built to Learn
Vesoma, a European startup with over 60 employees across Munich and Limassol, has emerged from stealth with plans to build humanoid robots that learn through physical interactions rather than relying solely on demonstrations. The company, led by former Google DeepMind research director Martin Riedmiller as chief AI officer, is building its first product called Vesoma 1 while also operating a walking prototype called Kyle.
Dynamic Creatures Emerges From Stealth With Boston Dynamics Backing to Build Expressive Guest-Facing Robots
Dynamic Creatures, a Boston-based startup backed by Boston Dynamics and several venture capital firms, has emerged from stealth to develop expressive, mobile 'living character robots' for the location-based entertainment and hospitality industry, powered by its proprietary SnowJay AI operating system. The company, co-founded by Boston Dynamics veterans Marc Theermann and Farbod Farshidian, plans to deploy autonomous bipedal and quadrupedal characters to theme parks, resorts, and retail venues, differentiating itself from warehouse-focused robotics competitors.
Gemini 4 Argon Takes Apollo From Human Video to Seven Straight Cornhole Shots
Google DeepMind demonstrated that its Gemini 4 Argon model can train Apptronik's Apollo 2 humanoid to sink seven consecutive cornhole shots by observing a human video, practicing in MuJoCo simulation, and refining the throw based on real-world feedback. The demo accompanies Argon's wider release, showcasing a video-to-simulation workflow that Google says reduces weeks of manual simulation modeling to hours. The article clarifies that Argon's reported 87% ERQA score measures embodied reasoning on multiple-choice questions, not robot task-completion performance.