- Written by Super User
- Category: Uncategorised
- Hits: 453
This website documents the progress of the Agribot projects led by the University of Strathclyde, which are intended to apply cutting-edge research from terrestrial and space robotics technology to the development of autonomous robotic systems for monitoring and treatment of crops.
Smart and automated farming practices are an essential technology for increasing yields, lowering production overheads, and maintaining the environment to ensure future productivity. Worldwide, it is now accepted that agricultural productivity will have to increase by 25% to allow limited arable land to meet a doubling of demand by 2050, and interest in agricultural robotics to allow real-time and accurate monitoring and response for crops has rapidly developed as a result.
The AgriRover prototype shown here was built this year by RAL Space and outfitted for autonomous agricultural monitoring duties by SMeSTech Strathclyde. It is a testbed for mobility, navigation, and sensing technologies to enable unsupervised and autonomous operation in monitoring farms. Some technologies are derived from the development of rovers and satellites for use in space, others are based on cutting-edge sensing and processing methods on Earth.
The AgriRover is powered for long periods by efficient lithium-ion batteries that can be recharged automatically at field stations, and navigates using a scanning laser rangefinder, hardware-accelerated stereo vision system, and short-range ultrasonic and inertial sensors. An extendable articulated arm will later hold sensors and allow haptic feedback to human operators as well as autonomous and semi-autonomous operation when gathering data in the field.
- Written by Mark Post
- Category: Rover platform
- Hits: 569
With the successful completion of the first funded Agribot projects, we have successfully completed U.K. field trials of the Agribot system and arm.
The navigation system has been tested successfully indoors, and also in outdoor field trials near Glasgow at the James Hutton Institute’s Hartwood Research Farm facility on 15, 17, and 18 March 2016. An additional field testing day was held at Rushyhill farm on 16 March 2016. The navigation system was tested each day and achieved success in mapping and obstacle detection, such as the example shown.
Mapping of a non flat surface. The map is correct at the beginning of the path. However, the inclination of the terrain is excessive at the end, due to accumulated odometry errors.
On the first day of trial, the autonomous navigation did not perform well. The robot could not locate itself within the field, and lost track of its position soon after initiating movement. The visual odometry technique was not sufficient, and the robot could not compute its own translation and rotation. The visual odometry algorithm has been subsequently re-configured for the tests on the following days.
The robot could be easily manoeuvred with a portable joystick. It mechanically performed properly on rough terrain and grassy areas. It showed its ability to overcome small obstacles, such as logs and molehills. However, its power was not enough to climb steeper uphill, especially after long exposure to cold winds decreases the battery output capacity.
A Grassy Field in Hartwood
On the following days, the navigation system parameters were changed. The visual odometry algorithm was allowed to be less demanding on the number of features require to localize the robot, and less precise. A camera supporting pole was added to improve the field of view and allow the system to gather more significant visual features. The robot could move autonomously and reach a new position without halting. A displacement error in the final pose was detected.