Search

This article presents a model predictive control based obstacle avoidance algorithm for autonomous ground vehicles in unstructured environments. The novelty of the algorithm is the simultaneous optimization of speed and steering without a priori knowledge about the obstacles. Obstacles are detected using a planar LIDAR sensor and a multi-phase optimal control problem is formulated to optimize the speed and steering commands within the detection range. Acceleration capability of the vehicle as a function of speed, and stability and handling concerns such as tire lift-off are taken into account as constraints in the optimization problem, whereas the cost function is formulated to navigate the vehicle as quickly as possible with smooth control commands. Thus, a safe and quick navigation is enabled without the need for a preloaded map of the environment. Simulation results show that the proposed algorithm is capable of navigating the vehicle through obstacle fields that cannot be cleared with steering control alone., Published in the Proceedings of ASME 2015 Dynamic Systems and Control Conference (DSCC 2015), held in Ohio, 28-30 October 2015.

A nonlinear model predictive control algorithm is developed for obstacle avoidance in high-speed, large-size autonomous ground vehicles (AGVs) that perceive the environment only through information provided by on-board sensors. The mission of the AGV is to move from its initial configuration to the goal configuration safely. The resulting trajectory should be collision-free and the AGV should be dynamically safe. As a starting point, the scenario where the vehicle moves on a flat surface at a constant speed is considered. The nonlinear MPC algorithm generates steering commands for completing the mission while enforcing safety constraints. The first safety constraint is avoiding obstacles. This is fulfilled by constraining the position of the AGV inside a safe region established from sensor data. The second safety constraint is ensuring dynamical safety. This is translated into avoiding single tire lift-off, which is implemented by limiting the steering angle within a range obtained using a 14 DoF vehicle dynamics model. At each sampling time, at least one multi-phase optimal control problem (OCP) is formulated and solved on-line. The safe region is partitioned into multiple sub-regions, which can then be specified without using piecewise functions. The fact that the optimal trajectory traverses the sub-regions sequentially and hence the position constraints are different from phase to phase makes the OCP multi-phase. The multi-phase OCP is transcribed into a nonlinear programming problem using the hp-pseudospectral method, and solved using the interior-point method. Simulations of an AGV approaching multiple obstacles show the effectiveness of the proposed algorithm., Sponsored in part by the Automotive Research Center, Ann Arbor, MI.

Unmanned ground vehicles (UGVs) are gaining importance and finding increased utility in both military and commercial applications. Although earlier UGV platforms were typically exclusively small ground robots, recent efforts started targeting passenger vehicle and larger size platforms. Due to their size and speed, these platforms have significantly different dynamics than small robots, and therefore the existing hazard avoidance algorithms, which were developed for small robots, may not deliver the desired performance. The goal of this paper is to present the first steps towards a model predictive control (MPC) based hazard avoidance algorithm for large UGVs that accounts for the vehicle dynamics through high fidelity models and uses only local information about the environment as provided by the onboard sensors. Specifically, the paper presents the MPC formulation for hazard avoidance using a light detection and ranging (LIDAR) sensor and applies it to a case study to investigate the impact of model fidelity on the performance of the algorithm, where performance is measured mainly by the time to reach the target point. Towards this end, the case study compares a 2 degrees-of-freedom (DoF) vehicle dynamics representation to a 14 DoF representation as the model used in MPC. The results show that the 2 DoF model can perform comparable to the 14 DoF model if the safe steering range is established using the 14 DoF model rather than the 2 DoF model itself. The conclusion is that high fidelity models are needed to push autonomous vehicles to their limits to increase their performance, but simulating the high fidelity models online within the MPC may not be as critical as using them to establish the safe control input limits., Submitted to Dynamic Systems and Control Conference 2013.

One of the main challenges of co-simulating hardware-in-the-loop systems in real-time over the Internet is the fidelity of the simulation. The dynamics of the Internet may significantly distort the dynamics of the network-integrated system. This paper presents the development of an iterative learning control based approach to improve fidelity of such networked system integration. Towards this end, a new metric for characterizing fidelity is proposed first, which, unlike some existing metrics, does not require knowledge about the reference dynamics (i.e., dynamics that would be observed, if the system was physically connected). Next, using this metric, the problem of improving fidelity is formulated as an iterative learning control problem. Finally, the proposed approach is utilized in a case study. The conclusion is that the proposed approach can significantly improve the fidelity in Internet-distributed hardware-in-the-loop simulation., Submitted to 2012 NDIA Ground Vehicle Systems Engineering and Technology Symposium, August 14-16, Troy, Michigan.

One of the main challenges of co-simulating hardware-in-the-loop systems in real-time over the Internet is the fidelity of the simulation. The dynamics of the Internet may significantly distort the dynamics of the network-integrated system. This paper presents the development of an iterative learning control based approach to improve fidelity of such networked system integration. Towards this end, a new metric for characterizing fidelity is proposed first, which, unlike some existing metrics, does not require knowledge about the reference dynamics (i.e., dynamics that would be observed, if the system was physically connected). Next, using this metric, the problem of improving fidelity is formulated as an iterative learning control problem. Finally, the proposed approach is illustrated on a purely simulation-based case study. The conclusion is that the proposed approach holds significant potential for achieving high fidelity levels., Paper submitted to Dynamic System and Control Conference.

Internet-distributed hardware-in-the-loop (IDHIL) simulation integrates geographically-dispersed HIL setups and fosters concurrent systems testing early in the design process. The degree to which an ID-HIL simulation loses fidelity relative to the single-location alternative is referred to as distortion. The literature showed that various factors affect distortion, e.g., delay, jitter, and loss. This paper shows that the coupling points, i.e., the particular points at which the system model is divided to enable distribution across the Internet, also affect distortion. The aim is to turn coupling point selection into a design decision that can be used to minimize distortion. To quantify distortion, a frequency-domain metric is proposed using a linear systems framework. This metric is then used to analyze how the choice of the coupling point affects distortion, leading to guidelines for selecting a coupling point that gives minimal distortion. The theory is demonstrated on a quarter-car model., Submitted for publication in a Special Issue of the International Journal of Vehicle Design. The original document contains color images.