The Use of Dynamic Sensing Strategies to Improve Detection for a Pepper Harvesting Robot*This research was supported by the European Commission (SWEEPER GA no. 644313) and by Ben-Gurion University of the Negev through the Helmsley Charitable Trust the Agricultural, Biological and Cognitive Robotics Initiative, the Marcus Endowment Fund, and the Rabbi W. Gunther Plaut Chair in Manufacturing Engineering. Academic Article uri icon

abstract

  • This paper presents the use of dynamic sensing strategies to improve detection results for a pepper harvesting robot. The algorithm decides if an additional viewpoint is needed and selects the best-fit viewpoint location from a predefined set of locations based on the predicted profitability of such an action. The suggestion of a possible additional viewpoint is based on image analysis for fruit and occlusion level detection, prediction of the expected number of additional targets sensed from that viewpoint, and final decision if choosing the additional viewpoint is beneficial. The developed heuristic was applied on 96 greenhouse images of 30 sweet peppers and resulted in up to 19% improved detection. The harvesting utility cost function decreased by up to 10% compared to the conventional single viewpoint strategy.

publication date

  • January 1, 2018