How Miso Robotics Automated Their Cooking with 3Laws
Miso Robotics were finding existing autonomous cooking solutions to be unreliable and costly. We provided a novel solution that has enabled them to decrease the need for human supervision, while delivering a consistent, high-quality food output.
- 27 times faster than the nominal motion planner
- 6.4x more resolution in time for trajectories, eliminating the need for local planning
- Functional in spaces as tight as 3cm of clearance
- Trajectories resulted in no collisions across 100 tests
Read the Full Study Here
Highlights:
- Collision-free Trajectories with Safety Guarantees: 3Laws Technology produces trajectories with minimal computation time and formal safety guarantees.
- Advanced Algorithms for Dynamic Environments: The system adjusts to changing conditions in real-time for highly accurate and efficient performance.
- Increased Automation and High-Quality Food Output: Provides increased automation to the cooking process and consistent, high-quality food output.
- Elimination of Manual Intervention: The system helped eliminate the need for manual intervention and provide a streamlined, automated cooking experience.
The Challenge: Difficult Motion Plans in a Dynamic Environment
Autonomous cooking is a challenge. Imagine a typical industrial kitchen in full swing: there is constant movement, and lots of it is very dynamic and intricate and improvised. Moreover, there are items and substances in play that present risks to safety and operational efficiency. No-one wants to spill hot oil, or damage a thousand-dollar knife.
This is why the automation of manual tasks such as pick-up, deep fry, and food dispensing has been such a puzzle. These tasks require the constant computation of motion plans in a dynamic environment. If the computation is off, it will inevitably lead to delays in scheduling and potentially over-cooked food.
3Laws wanted to improve how they automated their cooking, but existing solutions were falling short. They weren’t reliable, and their implementation came with decreased efficiency, increased costs, and reduced food quality.
The Solution: A Novel Safety Filtering Method from 3Laws
Miso Robotics came to 3Laws in search of a solution. We rose to the challenge by delivering a novel safety filtering method to manage and optimize the motion of cooking robots.
Our approach eliminates the need for constant re-planning in updated environments and provides robust safety guarantees for the resulting trajectory.
With this solution, the system is able to handle dynamic environments and adjust to changing conditions in real-time, resulting in highly accurate and efficient motions. Not only does this technology decrease the need for human supervision of the cooking process, but it also provides consistent, high-quality food output.
The Technical Details:
Our solution relies on control barrier functions (CBFs), which have been proven to provide effective means of enforcing safety on a wide variety of robotic systems, including robotic manipulators. The algorithm synthesizes safe velocity commands from existing, cached trajectories. It utilizes a full system characterization, without requiring explicit knowledge of the robot parameters. The method can handle complex obstacle representations and environments, making it practical for implementation. It is implemented in the MoveIt framework, which was used to construct planning scenes and compute signed distances.