Worker position tracking for safe navigation of autonomous orchard vehicles using active ranging

The harvesting of fresh market fruits is very labor intensive, since conventional mechanical and robotic harvesters have not yet replaced the judgment, dexterity and speed of farmworkers at a competing cost. Advanced labor aids such as autonomous fruit-transport robotic vehicles have been proposed to increase worker productivity in orchards by reducing or eliminating the time wasted for transporting harvested fruits. Much attention has been paid to autonomous navigation of orchard vehicles; yet, the issue of worker safety has not been addressed adequately. This paper constitutes a first step towards this goal. In particular it addresses the problem of human presence detection and tracking in orchards by using Ultra Wideband (UWB) radios that report the distance between the antennas of two radio units; combining four such distances from beacons on a vehicle and trilaterating, the coordinates of an antenna carried by a worker are calculated. In this study, GPS was used as the ground truth for evaluating the accuracy of the radio sensor system. Before testing, simulations were conducted to determine the optimal beacon configuration on the vehicle with respect to the localization error; the configuration with smallest error was selected for the experimental setup. Two sets of tests were conducted to evaluate the worker localization accuracy: one set measured static worker positions at points of varying distance and heading with respect to the vehicle, and the other set measured dynamic positions as the worker moved around vehicle in different size circles. The simulation and experiments both showed that the distance error was bigger as the distance from the vehicle grew. The experiment results indicated that 90% of the measurements had error less than 20 cm when the mobile sensor was about 1.5 m away from vehicle, and this value was 60 cm when the distance was about 5 m. These results indicate that the developed position-sensing system could be used for detecting and tracking the location of workers moving around automated orchard vehicles.