Publications

• Heiwolt, K., Mandil, W., Cielniak, G., and Hanheide, M. (2020). ‘Automated Topological Mapping for Agricultural Robots’, UKRAS20 Conference: “Robots into the real world” Proceedings, pp. 27-29, doi: 10.31256/Ze8Ex1V.
• Nazari, K., Mandil, W., Hanheide, M., and Ghalamzan Esfahani, A. (2021). ‘Tactile Dynamic Behaviour Prediction Based on Robot Action’, Annual Conference Towards Autonomous Robotic Systems (TAROS 2021), pp. 284-293, doi: 10.1007/978-3-030-89177-0_29.

Presentations

• UKRAS Conference (2020): Automated Topological Mapping for Agricultural Robots.
• Lincoln Conference on Intelligent Robots and Systems (2020): Investigation into Harvesting Soft Fruit Clusters.
• CRA Task-Informed Grasping Workshop – III (2021): Modelling soft fruit clusters for controlled harvesting. Watch here.
• AgriFoRwArdS CDT Annual Conference 2021, online (2021): Tactile prediction for controlled manipulation.

Other Activities and Outputs

• Took part in the AgriFoRwArdS Summer School 2021 resulting in co-authored presentation at the AgriFoRwArdS CDT Annual Conference 2021: Visual perception for harvesting grapes (in collaboration with Elijah Almanzor, Madeleine Darbyshire, Joshua Davy, Jerry Shi).
• Member of the discussion panel for Bob Fisher’s Keynote presentation at the AgriFoRwArdS CDT Annual Conference 2021, ‘The TrimBot2020 gardening robot and other agricultural robot issues’.
• Represented the CDT at the global climate change conference, COP26 in Glasgow (2021).
• Represented the CDT at the school outreach University of Lincoln British Science Week event (2022).

MSc Project

Investigation into Harvesting Soft Fruit Clusters

We propose an investigation into harvesting soft fruit clusters with robotics. Separating the problem into two distinct problems: motion planning to grasp the soft fruit (task 1); motion planning to release the soft fruit from the cluster (task 2). We will apply probabilistic movement primitives updated with model predictive control to task 1 and a probabilistic movement primitive architecture to task 2. We intend to test task 1 on strawberry clusters and task 2 on mushroom clusters.

PhD Project

Data-driven methods for detecting fruit and planning harvesting actions in dense cluster (DPFH)

Selective harvesting of soft fruit is a very challenging problem involving computer vision, motion planning, motion control, scheduling, and optimisation. State-of-the-art (SOTA) robotic system for selective harvesting of soft fruits are still far away from a human performance level, partially because the conventional planning cannot provide a feasible solution for the robot to reach-and-pick the soft fruit.   

DPFH aims to develop SOTA method for interactive motion planning and control for picking soft-fruits in a cluster where the robot needs to push the occluding pieces away to (i) better detect, segment and localise a ripe fruit and (ii) reach-and-pick the ripe fruit. The effectiveness of the developed methods in this project will be demonstrated in simulation environments, in the lab with toy strawberries and in the strawberry field.