Adnan Tahirovic

General Framework for Mobile Robot Navigation Using Passivity-Based MPC

This technical note proposes a novel navigation planner for mobile robots based on an adapted version of passivity-based nonlinear model predictive control. The proposed framework extends the convergent dynamic window approach and can be considered a generalized navigation planning technique able to include the high complex models required to describe the dynamics of vehicles moving outdoor on rough terrains. Several case studies are discussed to illustrate the usage of the framework.

An approximate of the cost-to-go map on rough terrains

The Roughness based Navigation Function (RbNF) is a numerical map that estimates the mobility measure (cost-to-go) from each terrain location toward the goal position. This paper compares the RbNF and the optimal cost-to-go map in terms of computational burden and solution quality. When the terrain is very large and obtaining the optimal cost-to-go map is computationally too expensive, the RbNF is shown to be able to compute an approximate solution much more quickly. As an application example, in which the RbNF is shown to be a powerful tool, the paper considers a Mars rover mission that finds the possible landing site using a mobility cost-to-go map constructed from a Mars terrain data.

A convergent solution to the multi-vehicle coverage problem

The paper presents a new solution to the multi-vehicle coverage problem. The proposed algorithm guarantees complete coverage and provides collaborative behaviors of vehicles, despite the fact that it does not explicitly exploit any computationally intensive optimization technique. The algorithm can deal with any mission domain, including regions with irregular shapes, multi-connected and disjoint regions. It gives reasonably good solutions even for partially connected multi-vehicle systems. The coverage problem for regions the shape of which change in time regardless the vehicle movement is also solved by the proposed algorithm.

Passivity-Based Model Predictive Control for Mobile Vehicle Motion Planning

Introduction.- PB/MPC Navigation Planner.- PB/MPC-RT Planner For Rough Terrains.- Conclusion.

PB/MPC Navigation Planner

In this chapter, a rather straightforward procedure is presented to obtain navigation algorithms for a broad class of vehicle models, based on an adapted version of the passivity-based nonlinear MPC examined in [1]. The proposed PB/MPC approach for navigation planning can be seen as a generalization of the well-known DWA developed in [2–4]. Similar to the navigation based on the MPC/CLF [5], the PB/MPC optimization setup guarantees the task completion, which means the vehicle is being able to reach the goal position. However, whereas in the MPC/CLF navigation framework a control action that decreases the Lyapunov function has to be found in advance, which is rather difficult if not impossible for complex vehicle models, the PB/MPC navigation framework gives directly the control action as a consequence of the passivity-based control. Therefore, the PB/MPC can be easily adapted to a variety of vehicle and terrain models providing a straightforward procedure for the navigation of wide range of vehicles.

A fast cost-to-go map approximation algorithm on known large scale rough terrains

Obtaining the optimal cost-to-go map for large scale rough terrains is computationally very expensive both in terms of duration and memory resources. A fast algorithm for approximation of the optimal cost-to-go map in terms of terrain traversability measures for path planning on known large scale rough terrains is developed. The results show that the majority of the cost-to-go map values, computed from every terrain location with respect to the goal location, are near-optimal. Unlike Dijkstra algorithm, the proposed algorithm has inherently parallel structure, and can be significantly speeded up depending on the number of used CPU cores.

Discussion on: ''Control and navigation in manoeuvres of formations of unmanned mobile vehicles''

Control of a multi-agent system has recently triggered a lot of research interest due to the inherently present problem complexity as well as various interesting applications that have become attractive in last few years, in which such an system can be utilized. The coverage problem, persistent monitoring, sensor networks, surveillance, formation control, highway and transportation systems are some of the possible application areas. Some relevant work can be found in the reference lists presented in [1,2]. Formation control has been a focus of the paper and it can be interpreted as a problem in which a set of mobile vehicles follows a reference trajectory while keeping a priori defined formation shape. Such a problem arises in some military applications as well as in applications like the one mentioned in the paper related to the airport snow shoveling. There are currently three different strategies adopted to solve the formation control problem: leader– follower approach, behavioral approach and virtual structure approach. In the paper, the leader–follower strategy is implemented based on a receding horizon control scheme (RHC). Some relevant work on the field of formation control can be found in the reference list presented in the paper. The main contributions of the paper are resulted by some inherent RHC framework features, providing a general approach which takes into account nonholonomic vehicles’ formation stabilization, a trajectory planning which guarantees task completion and a framework which handles dynamic changes in the environment. In addition, the RHC is enhanced by the possibility of the backward movement of the formations of unmanned mobile vehicles. The backward movement is accomplished by extending the leader– follower strategy with the idea of two virtual leaders which seems a reasonable way to solve this problem. Using an RHC scheme to control a formation of unmanned mobile vehicles allows for taking into account local measurements, which in turn makes the control capable of appropriately responding to dynamical changes in the environment during the mission of the mobile robots. The proposed RHC scheme includes mobility, environment and shape of the formation constraints to find suboptimal trajectories toward the target region. An additional term is also used in the objective function to ensure intervehicle collision avoidance. A capability of a backward movement of formations of unmanned mobile vehicles provides a wide range of possible applications in which the proposed control approach can be utilized. Such an additional feature makes the formations of unmanned mobile vehicles more flexible and capable for some complex applications such as the airport snow shoveling mentioned in the paper. However, the surveillance and the persistent monitoring coverage problem with a formation might be certainly enhanced by the backward movement feature. The proposed navigation framework extends the work [3] in that a detailed theoretical overview of the proposed method is covered jointly with some new experimental results. The paper additionally provides the analysis of the task completion, inherently implied by the convergence of the system into the desired target region. In the sequel, some different possible extensions of the

A side-scan sonar data-driven coverage planning and tracking framework

Abstract Mapping an unknown large-scale marine area by a side-scan sonar onboard a marine vehicle as quickly as possible is often of great importance. It is also important that a-priori unknown interesting parts of the area are scanned in more detail, i.e. with the removal of sonic shadows. In contrast to the standard overlap-all-sonar-ranges lawnmower pattern offline static coverage problem solution for side-scan sonar missions, here a novel online side-scan sonar data-driven coverage solution is proposed. The proposed coverage algorithm provides a coverage solution based on local information gain from side-scan sonar data. At the same time, the solution is generated in such a way that coverage path length/time is minimized while covering the same area as the standard lawnmower. Upper and lower bounds of the proposed algorithm’s improvement compared to the overlap-all-sonar-ranges lawnmower method are estimated analytically and validated through extensive mission parameters variation simulations. Simulation results show that this approach can cut down coverage path length/time significantly compared to the standard lawnmower method in most application cases.

All terrain vehicle path planning based on D* lite and MPC based planning paradigm in discrete space

This paper presents a multi stage technique for dealing with path planning problem on poorly traversable and partially unknown rough terrains. Traditionally, this problem is solved using D*-like algorithms, but by including a vehicle model, along with its constraints and the dynamics of the environment, this problem becomes very challenging. Another problem that requires a close attention is that the data measurements of the environment are usually discrete in nature, while current frameworks deal with mostly continuous data and systems. We propose solution based on fast D* lite algorithm for global path cost-to-go computation while employing MPC planning paradigm for solving constrained optimal control problem for the purpose of local planning. With this paradigm in use, both global, stationary, state of the environment, and local dynamics of the environment are taken into account in the near-optimal path planning.

RLS-based fault-tolerant tracking control of Multirotor Aerial Vehicles

This paper presents a fault-tolerant PD tracking system for Multirotor Aerial Vehicles (MAV) based on a novel Recursive Least Squares (RLS) Fault Detection and Isolation (FDI) algorithm utilized to diagnose propulsion system faults. As a test platform we investigate an octorotor model, including rigid body dynamics, the gyroscopic effect and motor dynamics. A hover configuration control is extended into an adaptive, fault-tolerant PD tracking controller. The approach is validated within a simulation study that includes a severe triple rotor fault scenario.