Sanctuary AI is a Canadian robotics company founded in 2018 by former executives from Kindred AI, a startup focused on using reinforcement learning for robotic manipulation. The company is headquartered in Vancouver, British Columbia, and is dedicated to building humanoid robots that can operate in human-centric environments. Its mission is to create the world's first human-like intelligence in general-purpose robots, with the goal of addressing labor shortages and transforming industries such as manufacturing, logistics, and healthcare.
Sanctuary AI's flagship product is Phoenix, a humanoid robot designed to perform tasks that are repetitive, dangerous, or difficult for humans. The company also develops a control system called Carbon, which is an AI software platform that enables the robot to learn and execute tasks through natural language instructions and demonstration. By combining advanced robotics with artificial intelligence, Sanctuary AI aims to deliver a workforce solution that can adapt to various environments and tasks.
Founding and Early Development
Sanctuary AI was founded in 2018 by Geordie Rose, Suzanne Gildert, Olivia Norton, and other former members of Kindred AI. Rose, who previously co-founded D-Wave Systems, a quantum computing company, brought a vision of creating humanoid robots with human-like cognitive abilities. The company initially operated in stealth mode, developing its core technologies before publicly announcing its mission in 2021.
In 2021, Sanctuary AI introduced its first-generation humanoid robot, Phoenix, which was designed to perform a range of tasks, from picking and placing objects to operating machinery. The robot was initially deployed in a pilot project with a major Canadian retailer, where it successfully completed tasks such as stocking shelves and bagging groceries. This early deployment demonstrated the potential of the technology and helped the company refine its approach.
Technology: Phoenix and Carbon
Phoenix is a humanoid robot that stands approximately 5 feet 7 inches tall and weighs around 155 pounds. It has two arms with seven degrees of freedom each, dexterous hands with 20 degrees of freedom, and a head with cameras and sensors for perception. The robot is designed to move safely in human environments, with a top speed of about 3 miles per hour and the ability to operate for up to 12 hours on a single charge.
The Carbon control system is the brain of Phoenix. It uses a combination of machine learning, deep learning, and neural networks to enable the robot to understand and execute tasks. Carbon can interpret natural language commands, learn from demonstrations, and generalize to new situations. It is built on a large language model architecture, similar to those used in generative AI systems, but adapted for physical action.
One of the key innovations of Carbon is its ability to learn tasks without extensive reprogramming. Operators can train Phoenix by showing it how to perform a task or by giving verbal instructions. The system then uses reinforcement learning to refine its skills over time. This approach reduces the need for specialized coding and makes the robot accessible to non-experts.
Deployment and Partnerships
Sanctuary AI has focused on deploying Phoenix in real-world settings to validate its technology. In 2022, the company partnered with a Canadian grocery chain to test the robot in a store, where it performed tasks such as stocking shelves and assisting customers. The pilot was successful, leading to further collaborations with other retailers and manufacturers.
In 2023, Sanctuary AI announced a partnership with a major automotive manufacturer to explore the use of Phoenix in assembly line operations. The robot was tested in tasks such as part handling and quality inspection. These deployments have provided valuable data for improving the robot's performance and reliability.
The company has also engaged with academic institutions and research organizations to advance the field of humanoid robotics. It has collaborated with the University of Toronto and Carnegie Mellon University on research projects related to robot learning and control.
Funding and Growth
Sanctuary AI has attracted significant investment from venture capital firms and strategic partners. In 2021, the company raised $30 million in a Series A round led by a prominent technology investor. In 2022, it secured an additional $75 million in Series B funding, bringing its total funding to over $100 million. The funds have been used to expand its engineering team, scale up production, and accelerate the development of Carbon.
In 2023, the company announced a partnership with Microsoft to integrate its AI capabilities with Azure cloud services. This collaboration aims to enhance the scalability and accessibility of Sanctuary AI's robots, allowing customers to deploy and manage fleets of Phoenix robots through the cloud.
Competitive Landscape
Sanctuary AI operates in a rapidly evolving field of humanoid robotics, competing with companies such as Figure AI, Tesla's Optimus, and Neuralink (though Neuralink focuses on brain-computer interfaces). Unlike some competitors that prioritize speed or specific tasks, Sanctuary AI emphasizes general-purpose capabilities and human-like intelligence.
The company's approach is distinct in its focus on natural language interaction and learning from demonstration. While other robots rely on pre-programmed routines or teleoperation, Phoenix is designed to understand and execute a wide variety of tasks with minimal human intervention. This positions Sanctuary AI as a leader in the quest for general-purpose robots.
Challenges and Future Directions
Despite its progress, Sanctuary AI faces significant challenges. Humanoid robots are complex and expensive to produce, and achieving reliable performance in unstructured environments remains difficult. The company must also address safety concerns and regulatory issues as robots become more integrated into workplaces.
Looking ahead, Sanctuary AI plans to expand its product line and improve the cognitive abilities of its robots. The company is investing in large language models and transformer architectures to enhance Carbon's understanding and reasoning. It also aims to reduce the cost of Phoenix to make it accessible to small and medium-sized businesses.
In 2024, Sanctuary AI announced a roadmap to deploy thousands of robots across various industries within the next five years. The company is also exploring applications in healthcare, where robots could assist with patient care and rehabilitation.
Ethical and Societal Implications
Sanctuary AI recognizes the broader implications of humanoid robots on society. The company has established an ethics advisory board to guide its development and deployment practices. It emphasizes the importance of creating robots that augment human workers rather than replace them, and it advocates for policies that support workforce transition.
The company also addresses concerns about privacy and data security, ensuring that its robots are designed with safeguards to protect sensitive information. As humanoid robots become more common, Sanctuary AI aims to be a responsible leader in the field, promoting transparency and accountability.
Conclusion
Sanctuary AI is at the forefront of developing humanoid robots with human-like intelligence. Through its Phoenix robot and Carbon control system, the company is pushing the boundaries of what is possible in robotics and AI. With substantial funding, strategic partnerships, and a clear vision, Sanctuary AI is poised to play a major role in shaping the future of work and human-robot interaction.