Evaluation of Warning Methods for Remotely Supervised Autonomous Agricultural Machines

Area of Science:

• Agricultural Engineering
• Human-Machine Interaction
• Automation Systems

Background:

• Autonomous agricultural machinery requires human supervision for safe operation.
• Timely recognition of machine malfunctions by supervisors is crucial for intervention.
• Current warning systems utilize visual, auditory, and tactile senses with varying effectiveness.

Purpose of the Study:

• To determine the most effective warning methods for remote supervisors of autonomous agricultural machines.
• To compare the efficacy of different sensory cues (visual, auditory, tactile) in emergency situations.
• To evaluate warning method suitability across various remote supervision scenarios.

Main Methods:

• Participants interacted with a simulated autonomous sprayer system.
• Seven distinct warning methods, including sensory cue combinations, were tested.
• Four remote supervision scenarios were simulated: in-field, close-to-field, farm office, and outside farmland.

Main Results:

• A combination of tactile and visual warning methods proved most effective for in-field and close-to-field supervision.
• No definitive best warning methods were identified for supervisors located at the farm office or further away.
• Effectiveness varied significantly based on the sensory modality and supervision proximity.

Conclusions:

• Visual and tactile warnings are essential for supervisors monitoring autonomous agricultural machines in near-field operations.
• Improved warning systems can enhance remote supervisor performance and reduce operational risks.
• Further research is needed to optimize warning strategies for distant supervision scenarios.