The MeSAPro project provides the necessary requirements for using Robotics and Autonomous Systems (RAS) in soft fruit production to assure system safety during Human-Robot Interaction (HRI).

The challenge

Safety is a recognised problem in agricultural work. Agricultural workers make up only 1% of the UK workforce, yet suffer 20% of all workplace fatalities, and 4% of farmworkers suffer non-fatal accidents each year.

To enhance safety in agriculture, the University of Lincoln and SAGA Robotics have developed fleets of autonomous soft-fruit production robots that undertake several tasks from monitoring and forecasting to harvesting and transportation in real-world fruit farm environments.

The research

This project complements the team’s existing work on the functionality and technical capabilities of the robots by:

• defining safety requirements for the sense-understand-decide-act components of a soft-fruit production RAS.
• developing methods to detect deviations from safe behaviour and ways to mitigate the effect of such deviations.
• formally verifying sensing, understanding and deciding components of a soft-fruit production RAS.

The progress

The team has developed its descriptions of four scenarios for the deployment of robots in soft-fruit harvesting. They have completed a hazard analysis for each scenario, identifying the critical points at which safety could be compromised, and followed these up by identifying how these hazards can be mitigated.

They are currently implementing these mitigations and integrating them into the existing software stack that is used on the Thorvald robots that are employed commercially. In parallel, the team has begun the formal verification of the safety system. Some initial results on both the hazard analysis and the formal verification have been written up for publication.

This is a demo of the team's second version of the proposed safety system architecture for the MeSAPro project. New features include the human detection system. The system is now running with the Thorvald navigation pipeline.

You can see a previous demo video on YouTube.

Papers and presentations

• Khalid, M., Guevara, L., Hanheide, M., Parsons, S. "Assuring autonomy of robots in soft fruit production", 4th UK-RAS Conference for PhD Students & Early-Career Researchers, 2021.
• Guevara, L., Khalid, M., Hanheide, M., and Parsons, S. "Assessing the probability of human injury during UV-C treatment of crops by robots", 4th UK-RAS Conference for PhD Students & Early-Career Researchers, 2021.

Project partners

• Lincoln Centre for Autonomous Systems, University of Lincoln
• SAGA Robotics