Reconfigurable Collaborative Agri-Robots (Recoaro)
In rugged, mountainous regions, like South Tyrol, many agricultural tasks are performed manually, which can lead to uncertainty in food production and cost, seasonal labor shortages, increased risk of fatal accidents, and adverse long-term health effects on farm workers. Alpine agricultural environments are complex and unstructured, limiting the use of autonomous systems. Further, most Alpine farms are relatively small and cannot afford the cost of several machines that are specialized for only one or two tasks.
Aim: develop new methods that can enable agricultural robotic systems for Alpine farms to be modular, reconfigurable and semi-autonomous, combining automation with human input to create a collaborative, flexible, human-robot team.
– kinematic/dynamic modelling
– automatic control
– trajectory planning
– shared human-robot control of modular reconfigurable robotic (MRR) platforms.
Project objectives address the following open problems:
1) How to automatically adapt dynamic/kinematic models to capture the changing characteristics of MRRs as they are reconfigured? We will develop systematic methods to update the kinematics/dynamics using data stored in each module, accounting for model uncertainties using interval arithmetic.
2) How can model-based controllers for mobile MRRs be adapted as the underlying dynamic/kinematic models of the MRR change? We will develop new centralized control schemes for mobile MRRs using interval arithmetic to ensure robust stability and tracking performance.
3) MRRs can become highly redundant, complicating trajectory planning, but simplifying optimization. How can low-level problems (self-collision, joint-limit avoidance) and high-level problems (planning time/energy optimal trajectories) be solved as a task is executed? We will study how to extend task-prioritized redundancy resolution to MRRs for trajectory planning.
4) Few studies address the stability of shared human-robot control systems. We will explore if homogeneously rescaled shared control laws can be used to ensure the Lyapunov stability of MRRs.
5) In shared trajectory tracking control a human may periodically intervene and move the MRR towards a new trajectory. How should trajectory planning be handled when the human returns control to the robot? We will examine how to combine global and local trajectory planning for shared control using homogenous transformations.
6) Experimental validation of the methods developed will be conducted at the UniBZ FIeld Robotics South Tyrol (FIRST) Lab and at Fraunhofer Italia’s (FhI’s) Area for REsearch and iNnovative Applications (ARENA), both of which are located in the NOI Techpark. The experiments will simulate three common farming tasks: spraying, inspection and harvesting.
SPRUCE-ROBOT: Smart PRUning and Climbing treEs ROBOT
The production of high value-added wood requires proper pruning during the life of a tree so that its trunk has a cylindrical shape and does not contain knots. In order to accomplish this, usually either the development of branches is mitigated, or branches are removed before they dry so that knots do not form along the basal part of the trunk. The second strategy, the most interesting, can be achieved by “Green pruning operations”, which are currently conducted by human operators who remove branches using ladders, crampons, manual saws or light chainsaws in expensive and time-consuming operations.
This applied interdisciplinary research project aims to apply mechatronic approaches for automating an important forestry activity, like tree pruning, so that it is safer and more efficient.
Starting with an extensive literature review, and thanks to the cooperation of our external partner, the Azienda Demanio Provinciale, the main requirements for a novel robotic climbing and pruning prototype will be defined for use in spruce forests.
The interdisciplinary team will study, design and develop field robotics technologies and practices that can assist the forestry industry to produce timber more efficiently, more safely and with less waste. In particular, a modular and adaptable robotic tree climbing prototype, which can be used by UniBZ to develop and test new pruning technologies, will be studied, designed and developed.
Different challenging problems will be tackled, like the design of:
(1) a proper climbing strategy that allows rapid climbing and with minimal bark damage;
(2) an effective sensing system able to localize branches and their geometrical features, as well as estimate the cutting forces required;
(3) an adjustable and smart cutting system. Moreover, path and trajectory planning algorithms, as well as decision-making and task
planning strategies, will be developed to properly control the robotic platform.
Extensive tests will be performed in both controlled laboratory conditions and in-field to validate the proposed system.