Marta Niccolini

Near optimal swarm deployment using Descriptor Functions

The article describes a novel approach for deployment of a swarm of heterogeneous autonomous vehicles. Each vehicle is treated as the agent of a network, which cooperates in order to cover a given area according to its own capabilities. A general framework is introduced, which aims at providing tools for solving a large class of coordination problems. The capabilities of each agent are modeled with Descriptor Functions; the sum of these functions constitutes the swarms descriptor. The goal of the swarm is to match a desired descriptor, by minimizing an appropriate cost functional. A control law is proposed, which is capable of driving the agents towards the achievement of the goal. The existence of local minima and of a global minimum is discussed. Theoretical results on the existence of the global minimum are given. The Area Coverage problem is selected as a preliminary test for the algorithm. Simulation results show the effectiveness of the proposed approach.

Human-Swarm Interface for Abstraction Based Control

Decentralized control of a large number of agents acting as a single swarm is an active fleld of research. More recently the problem of how to instruct the swarm to perform certain tasks, and in particular, how to interface the swarm to a human operator has arisen. This paper describes a Human Swarm Interface under development at University of Pisa which aims at providing the human operator with an easy to use display console and an input device which allows him/her to intuitively control and monitor a swarm position and shape over time. The swarm position and shape is deflned using an abstraction based approach; each single swarm agent estimates the swarm state via a distributed consensus algorithm and moves according to a steepest descent algorithm toward minimization of the error between human operator command and swarm state. The paper presents some new developments of the DEAS testbed as well, where the proposed interface is implemented and tested, which face the issues of inter agent synchronization and con∞ict-free wireless communications. Simulation and experimental results are shown.

Cooperative Control for Multiple Autonomous Vehicles Using Descriptor Functions

The paper presents a novel methodology for the control management of a swarm of autonomous vehicles. The vehicles, or agents, may have different skills, and be employed for different missions. The methodology is based on the definition of descriptor functions that model the capabilities of the single agent and each task or mission. The swarm motion is controlled by minimizing a suitable norm of the error between agents’ descriptor functions and other descriptor functions which models the entire mission. The validity of the proposed technique is tested via numerical simulation, using different task assignment scenarios.

Multiple UAV Task Assignment using Descriptor Functions

Abstract The paper presents a preliminary application of a new methodology to the problem of task assignment, for a large number of unmanned vehicles operating in a cooperative and coordinated fashion. The proposed methodology uses the general concept of Descriptor Functions, to establish a relationship between tasks, their priority, and agents or groups of agents forming a large interconnected system. A bio-inspired model of the division of labor in colonies of social agents is used to derive a set of rules at the individual level that gives rise to agent specialization. Preliminary simulations show interesting results, and the potential of the proposed technique.

Towards Real-Time Aircraft Simulation with the MPI Motion Simulator

The paper describes the recent advancements gained on the MPI motion simulator project. The aim of this project is the use of an a nthropomorphic robot as actuation system for a motion platform intended for real time flight simulation. Almost all commercially available motion platforms rely on the so called St ewart platform, that is a 6-DOF platform that can bear high payloads and can achieve high accelerations. On the other hand an anthropomorphic manipulator offers a larger range of motion and higher dexterity, that let envisage this novel motion simulator as a viable an d superior alternative [1,2]. The paper addresses the use of a new inverse kinematics algor ithm capable of keeping joint velocities and accelerations within their limits. Preliminary experimental results performed using the proposed algorithm along with possible further improvements are discussed.

New interaction metaphors to control a hydraulic working machine's arm

In the last decade, teleoperated hydraulic machines have become a reality in post-disaster scenarios. Although Human Machine Interaction technologies had huge advances, these technologies are not always applied successfully, especially in fields in which automation or semi-automation have difficulties to make their way, i.e. unstructured scenarios. This paper aims to introduce new interaction metaphors to improve operation of modern teleoperated hydraulic machines. Three new prototypes of Human Machine Interfaces are presented: a wireless portable joystick, a wearable hand motion recognition system and two facing haptic interfaces. The interfaces are evaluated in a typical pick and place scenario, developed with the aid of Virtual Reality. The results are discussed from an ergonomic and task performance point of view. The wireless portable joystick and the haptic interface produced similar outcomes and they are both usable for the control, whereas the wearable hand motion control proved to be unsuitable the task.

A Novel Wearable Haptic Controller for Teleoperating Robotic Platforms

This letter presents a novel wearable haptic controller (WHC) system suitable for teleoperation of demolition machines and robotic platforms. With regard to existing operator controller unit composed by passive joysticks, the WHC has been designed to provide force feedback to the user, hence improving the user performance during the teleoperation of different robotic platforms and the interaction with the environment on the remote construction sites. The haptic feedback is provided through two compact parallel kinematic (CPK) interfaces that will be presented within the paper. The CPK implements a novel variant of the Delta kinematics which allows minimizing the radial encumbrance while preserving same operational workspace. In addition, we propose a new interaction modality that provides users the feeling of directly maneuvering the end-effector of the demolition machine. Finally, the architecture of the proposed system is presented and the results of some preliminary evaluation tests are discussed. The experiments have been performed in simulated environments and on a real machine.

Cartesian-Space Motion Planning for Autonomous Construction Machines

Nowadays, the construction industry is probably the least productive and most dangerous among the various industry sectors. Given this scenario, it is quite clear that the introduction of Autonomous Construction Machines (ACMs) could represent a great opportunity to improve both productivity and safety. To this purpose, a fundamental problem that has to be tackled is trajectory planning. In the last 15 years, several sample-based algorithms have been proposed, that relies on Joint-Space sampling. Unfortunately, this feature often results in trajectories that are quite counterintuitive from the point of view of a human being. In this work we propose “cart-RRT”, a Cartesian-Space randomized algorithm that improves the intuitiveness of the output trajectory, while ensuring both its safety (in terms of collision avoidance) and its

Decentralized Control of Swarms with Collision Avoidance Implications

The paper reviews some approaches to the decentralized control of a swarm of unmanned vehicles, and then proposes a new algorithm capable of managing collisions between vehicles and with obstacles. The swarm goal is to achieve a desired shape and position in space, formalized using an abstraction based approach. Formation statistics are defined in analogy with phisical bodies: center of mass position and inertia moments. Each agent elaborates its own estimate of these variables using a consensus algorithm capable of tracking a ramp reference, in order to reduce tracking errors. Gyroscopic and damping terms are added to the control law in order to avoid collision between vehicles and with obstacles. The obstacle avoidance terms appear in the control law only in the presence of obstacles or nearby vehicles; thus system dynamics change during the system evolution. This behavior was modeled as a hybrid system and proof of stability is given, under mild conditions, using the Common-Lyapunov function approach. The proposed methodology is validated through extensive numerical simulation.

Experimental Evaluation of Decentralized Swarm Control Laws

The paper proposes an experimental setup designed for testing of decentralized algorithms for a swarm of unmanned vehicles. The agents of the swarm are low-cost electric radio controlled (RC) cars equipped with a GPS receiver and the electronics needed to communicate with other agents and a remote station. A remotely piloted system has been setup to allow on the ∞y tuning of the controllers, and to perform hardware in the loop simulations; this testbed can be used to create ad-hoc scenarios, to simulate particular system states and to address the implications due to non-ideal vehicles. A kinematic model of the single agent was derived in order to allow the design of low level controllers and to simulate systems of non-holonomic agents. The formation keeping problem is used to prove the efiectiveness of our experimental setup. The swarm goal is to achieve a desired shape and desired position in space. This is formalized using an abstraction based approach, which deflnes formation statistics in analogy with physical bodies: center of mass position and inertia moments. Each agent elaborates its own estimate of these variables using dynamic consensus algorithms. Both simulated and experimental results obtained with the testbed highlighted the possibility of limit-cycle like instability of the formation of agents. A decentralized guidance system is proposed to suppress this instability. Numerical and experimental results are presented.