Hello Robot Deploys $29,950 Stretch 4 for Home Assistive Care
Hello Robot's Stretch 4 mobile manipulation platform is now deploying to home environments with a dual-processor architecture and human-in-the-loop control.
California-based startup Hello Robot has transitioned its Stretch 4 mobile manipulation platform from laboratory testing into active home deployment. Priced at $29,950, the initial production run of the fourth-generation hardware has sold out. The open-source system targets researchers, data centers, and assistive technology developers building physical AI applications.
Hardware Architecture
Stretch 4 abandons the complex bipedal architecture pursued by systems like the Isaac GR00T platform in favor of a 46 kg pragmatic design. The system utilizes a slender telescoping arm mounted on a new holonomic omnibase. Three 8-inch wheels provide omnidirectional motion capable of navigating household thresholds and rugs.
Compute is divided across a dual-processor architecture. An Intel Core Ultra 5 225H handles primary operations, while an auxiliary NVIDIA Jetson Orin NX accelerates local vision and AI workloads.
The sensing suite includes two hemispherical Hesal J128 3D LiDARs providing a 360-degree by 187-degree field of view. Visual input relies on dual global-shutter 2.3MP RGB cameras and a single high-resolution 12MP RGB camera. Six dedicated laser line sensors handle floor hazard detection to navigate cords and drop-offs. The arm extends to a 55 cm reach with a 2.5 kg extended payload, powered by a 512Wh LiFePO4 quick-swap battery that yields four to eight hours of runtime.
Assistive Care Integration
Recent home deployments prioritize a human-in-the-loop control philosophy over full autonomy. Users maintain direct oversight through simple interfaces. In active case studies, paralyzed users operate the robot via voice commands to a smartphone application to handle daily tasks like drinking and retrieving items.
Researchers at the University of New Hampshire are testing the platform for dementia support. The robot provides automated reminders for hydration and nutrition while monitoring environmental safety for elderly couples. The lightweight frame minimizes the safety risks associated with heavier humanoid robots falling in residential settings.
If you are fine-tuning robotics models, Stretch 4 offers a standardized hardware baseline for accumulating real-world operating hours. The current deployment strategy focuses on gathering this specific home interaction data to train more capable assistive and physical AI models.
Get Insanely Good at AI
The book for developers who want to understand how AI actually works. LLMs, prompt engineering, RAG, AI agents, and production systems.
Keep Reading
How to Control Agent Tool Execution via Genkit Middleware
Learn how to use Google's new Genkit Middleware to intercept model calls, implement human-in-the-loop tool approvals, and handle transient API failures.
$8.2M Seed Backs Human Archive's Gig-Worker Robotics Dataset
Human Archive has raised $8.2 million to build a multimodal robotics dataset by paying Indian gig workers $1 per hour to record physical service tasks.
Gemini Enterprise Gains Agentic RAG for Multi-Hop AI Queries
Google Research launched a multi-agent retrieval framework in Gemini Enterprise that dynamically searches across data islands and verifies context accuracy.
Thousand Token Wood Runs a 5-Agent Economy on Qwen2.5-3B
Developed for Hugging Face's Build Small Hackathon, the Thousand Token Wood simulation uses a 3-billion-parameter model to drive a real-time agent economy.
iMessage Supports First Third-Party AI Agent With Poke Approval
Apple has approved Poke as the first AI agent for Messages for Business, allowing users to execute tasks and interact with an AI assistant directly in iMessage.