Project

CROPS – Intelligent robot systems for greenhouse horticulture

The European research project Clever Robots for Crops (CROPS) to develop robotics in horticulture and forestry has been completed. The project coordinated by Wageningen University & Research Greenhouse Horticulture unit and co-financed by the Product Board for Horticulture has led, among other things, to the development of the bell pepper harvesting robot.

The four-year research project with thirteen partners from ten different countries has resulted in a universal robotic platform with intelligent sensing and manipulation for sustainable production and harvesting high-value crops. Demonstration robots for picking peppers, apples, and grapes, precision spraying of pesticides, and obstacle-avoidance sensors for forestry machinery were developed. All these robots use the same type of modular system and software architecture, which makes it easy to use, for example, a different gripper or camera system. The CROPS project was followed up by the SWEEPER project.

Project highlights

  • Modular smart platform: CROPS developed scientific know-how for a highly configurable, modular and smart carrier-platform that includes modular parallel manipulators and intelligent tools (sensors, algorithms, sprayers, and grippers) that can be easily installed onto the carrier and that are capable of adapting to new tasks and conditions.
  • Technological demonstrators: Several technological demonstrators were developed for high value crops like greenhouse vegetables, fruits in orchards, and grapes for premium wines. The CROPS robotic platform is capable of site-specific spraying (targets spray only towards foliage and selective targets) and selective harvesting of fruit (detects the fruit, determines its ripeness, moves towards the fruit, grasps it and softly detaches it).
  • Detection and classification: Another objective of CROPS was to develop techniques for reliable detection and classification of obstacles and other objects to enable successful autonomous navigation and operation in plantations and forests. The agricultural and forestry applications share many research areas, primarily regarding sensing and learning capabilities.

Results

In 2013, a number of laboratory experiments were carried out. The main objective of the laboratory experiments was to obtain data on the performance and accuracy of the various subsystems. During these trials, 189 of the 194 fruits could be detected (97%), 167 fruits could be approached (86% of all fruits) and 154 picked (79% of all fruits). The main reasons for failed picking operations were manipulator problems (failing brakes) and inaccurate detection of fruit positions. After conducting the laboratory experiments, a list of suggested improvements was discussed with the consortium.

With the first functional bell pepper harvesting robot working in a realistic environment, this project reached an important milestone

For the greenhouse experiments, a number of modules were replaced by a newer version, such as the final manipulator, a new lighting unit for the cameras, and new prototypes of the two end-effectors. Between May and July 2014, greenhouse experiments with the integrated bell pepper harvesting robot in a crop with red bell peppers were carried out at a Dutch grower. During these experiments, the feasibility of autonomous harvesting of peppers was proven. The experiments yielded a wealth of information and a number of issues were identified that caused the limited performance. The percentage of fruit harvested flawlessly and without damage in an unadapted field crop was only between 2% and 6%.

After simplifying the crop by removing fruit clusters and removing leaves, which limited the sensors' unobstructed view of the fruits, the success rate was improved to 33%. The average turnaround time to pick a fruit was 94 seconds. Picking success and lead time are still insufficient in practice, but with the first working bell pepper harvesting robot in a realistic environment, an important milestone was reached in this project. Experiments were also carried out with a method newly developed in this project, which takes the position of the main stem of the plant into account when gripping the fruit. By using this method, the damage to the stem by the robot decreased.

The bell pepper harvesting robot was demonstrated several times during the project. In July 2013, a first version of the fully integrated system with the first prototype of the manipulator was demonstrated to the members of the supervisory committee (BCO) in the WUR laboratory. In addition, the robot was demonstrated to visitor groups at various stages of development. The demonstration of the system in the greenhouse for the BCO took place on July 3, 2014.

The robot was also shown to the scientific public during the closing CROPS workshop on July 8, 2014 during the Agricultural Engineering Conference (AgEng 2014) in Zurich, Switzerland. The robot was also demonstrated live to a wide audience at a special organized information and demonstration day in the Netherlands on September 4, 2014. Several video clips of the bell pepper harvesting robot in the greenhouse were produced and can be downloaded from the video section of the project website. The project-related PhD thesis by C.W. Bac “Improving obstacle awareness for robotic harvesting of sweet-pepper” was published in January 2015.