Planning and Control of Uncertain Cooperative Mobile Manipulator-Endowed Systems under Temporal Logic Tasks
Time: Fri 2020-05-29 15.00
Location: Zoom (English)
Subject area: Electrical Engineering
Doctoral student: Christos Verginis , Reglerteknik
Opponent: Magnus Egerstedt,
Supervisor: Dimos Dimarogonas,
Control and planning of multi-agent systems is an active and increasingly studied topic of research, with many practical applications such as rescue missions, security, surveillance, and transportation. This thesis addresses the planning and control of multi-agent systems under temporal logic tasks. The considered systems concern complex, robotic, manipulator-endowed systems, which can coordinate in order to execute complicated tasks, including object manipulation/transportation. Motivated by real life scenarios, we take into account high-order dynamics subject to model uncertainties and unknown disturbances. Our approach is based on the integration of tools from the areas of multi-agent systems, intelligent control theory, cooperative object manipulation, discrete abstraction design of multi-agent-object systems, and formal verification.
The first part of the thesis is devoted to the design of continuous control protocols for the cooperative object manipulation/transportation by multiple robotic agents, and the relation of rigid cooperative manipulation schemes to multi-agent formation. We propose first a variety of centralized and decentralized control algorithms that do not employ force/torque information at the contact points and take into account both cases of rigid and rolling grasping points, dynamic uncertainties in the object’s and agents’ model, and potential constraint satisfaction, such as obstacle avoidance and input saturation. Next, we tackle the problem of robust formation control for a class of multi-agent systems and we analyze the relation between formation rigidity theory and rigid cooperative manipulation.
In the second part of the thesis, we develop control schemes for the continuous coordination of multi-agent complex systems with uncertain dynamics. We first study the motion planning problem and propose novel adaptive control schemes for the collision-free navigation of single- and multi-agent spherical systems in obstacle-cluttered environments. Next, we focus on the leader-follower coordination problem of spherical multi-agent systems. More specifically, we design a robust adaptive decentralized control scheme for the successful navigation of a designated leader to a predefined point, while guaranteeing collision avoidance and connectivity maintenance properties. Finally, we design a closed-form robust barrier function-based control protocol for the collision avoidance of multiple 3D ellipsoidal agents.
The third part of the thesis is focused on the planning and control of multi-agent and multi-agent-object systems subject to complex tasks expressed as temporal logic formulas. We tackle first the case of local independent tasks for multi-agent systems, and by using previous results on multi-agent constrained navigation, we design a discrete abstraction of the agents’ motion in the workspace and synthesize decentralized control policies that satisfy the agents’ specifications. Next, in addition to the robotic agents,we take into account complex tasks to be satisfied by unactuated objects. We design a discrete abstraction that simulates the behavior of the agents and the objects in the workspace and we synthesize controllers for the agents that take into account both theirs and the objects’ specifications.
The fourth and final part of the thesis focuses on several extension schemes for single-agent setups. Firstly, we consider the problem of single-agent motion planning under timed temporal tasks in an obstacle-cluttered environment. Using previous results on collision-free timed navigation, we develop a novel control policy that guarantees satisfaction of the agent’s timed tasks as well as asymptotic optimality with respect to energy resources. Secondly, we tackle the motion planning problem for high-dimensional complex systems with uncertain dynamics in obstacle-cluttered environments. We integrate intelligent control techniques with sampling-based motion planning algorithms to guarantee the safe navigation of the system to a predefined goal, while compensating for the model inaccuracies. Finally, we develop a novel control protocol that achieves asymptotic reference tracking for an unknown control affine system, while respecting at the same time funnel constraints.