View PDF

Abstract:Rising labor costs and narrow treatment windows have made teleoperated robots a proposed tool for vineyard scouting and treatment, but simulation-based comparisons of human and robotic performance are sensitive to modeling choices that can favor one agent before any trial runs: whether both can record a target's location, whether both are routed efficiently, and whether travel speeds are assessed consistently. This paper simulates a human operator against a teleoperated robot on two vineyard tasks, disease detection-and-treatment and yield-zone survey, holding both agents to matched location-marking and routing standards, using a routing strategy proven optimal for this topology, and modeling teleoperation-specific effects. Once information access and routing are held equal, task-time outcomes are governed largely by relative sustained travel speed: at the speeds modeled (1.4 m/s human, 0.9 m/s UGV), the human completes both tasks 43-48 percent faster, a gap persisting across a simulated eight-cycle season, while at speed parity the agents differ by only 9-12 percent, and every other parameter shifts the result by under 4 points. These results indicate the comparison is primarily a question of platform speed, and that the case for teleoperated robots should rest on labor availability, endurance, consistency, or hazard avoidance rather than completion time.

Submission history

From: Daniel Udekwe [view email]
[v1] Sat, 5 Jul 2025 21:36:02 UTC (2,991 KB)
[v2] Wed, 26 Aug 2026 14:43:12 UTC (889 KB)