Flexible AI-controlled picking robot with a higher degree of autonomy
| Reference number | |
| Coordinator | Superintelligence Computing Systems SICSAI AB |
| Funding from Vinnova | SEK 6 500 000 |
| Project duration | October 2023 - October 2025 |
| Status | Ongoing |
| Venture | Advanced digitalization - Enabling technologies |
| Call | Advanced and innovative digitalization 2023 - call two |
Purpose and goal
We are developing the next generation AI controlled industrial pick-&-place robot with a higher degree of autonomy and flexibility to increase the percentage of automatic picking. We meet the strategic demand for automated picking in an unstructured warehouse with many items and high turnover of items, an important and unsolved automation problem. Automatic picking provides productivity gains of 5x compared to manual. Today´s AI is insufficient for modeling complex objects, learning new objects online, planning and learning an efficient picking strategy. Our new AI solves this.
Expected effects and result
The project goal is for the consortium members to jointly develop the components and system solutions required to realize the next generation AI-controlled industrial pick-and-place robot with a new type of AI architecture that provides a higher degree of autonomy and flexibility, with the possibility to: Increase the share of automatic picking within warehouse logistics Increase picking accuracy Increase picking speed Increase supply chain integration Increase human-system interaction based on natural language
Planned approach and implementation
The development is iterative and goes from solving simpler problems in simplified environments to greater complexity, from simulation in a virtual world to tests in the physical world, from lab tests to tests with the end customer in an industrial environment. Total project time is 24 months with a planned project start in October 2023. We work agile with continuous delivery of improved functionality in a number of different tracks or work packages. We build end-to-end versions of the system early with testing in physical environment for fast feedback loop.