Bai Li

A unified motion planning method for parking an autonomous vehicle in the presence of irregularly placed obstacles

This paper proposes a motion planner for autonomous parking. Compared to the prevailing and emerging studies that handle specific or regular parking scenarios only, our method describes various kinds of parking cases in a unified way regardless they are regular parking scenarios (e.g., parallel, perpendicular or echelon parking cases) or not. First, we formulate a time-optimal dynamic optimization problem with vehicle kinematics, collision-avoidance conditions and mechanical constraints strictly described. Thereafter, an interior-point simultaneous approach is introduced to solve that formulated dynamic optimization problem. Simulation results validate that our proposed motion planning method can tackle general parking scenarios. The tested parking scenarios in this paper can be regarded as benchmark cases to evaluate the efficiency of methods that may emerge in the future. Our established dynamic optimization problem is an open and unified framework, where other complicated user-specific constraints/optimization criteria can be handled without additional difficulty, provided that they are expressed through inequalities/polynomial explicitly. This proposed motion planner may be suitable for the next-generation intelligent parking-garage system.

Precise trajectory optimization for articulated wheeled vehicles in cluttered environments

An articulated vehicle trajectory optimizer is proposed.Constraint violations between every two adjacent collocation points are considered.A simple but accurate collision avoidance judgment is utilized.Large-scale constraints are incorporated into the optimization objective.Various optimization objectives can be handled by our trajectory optimizer uniformly. Trajectory planning refers to planning a time-dependent path connecting the initial and final configurations with some special constraints simultaneously considered. It is a critical aspect in autonomously driving an articulated vehicle. In this paper, trajectory planning is formulated as a dynamic optimization problem that contains kinematic differential equations, mechanical/environmental constraints, boundary conditions and an optimization objective. The prevailing numerical methods for solving the formulated dynamic optimization problem commonly disregard the constraint satisfactions between every two adjacent discretized mesh points, thus resulting in failure when the planned motions are actually implemented. As a remedy for this limitation, the concept of minute mesh grid is proposed, which improves the constraint satisfactions between adjacent rough mesh points. On the basis of accurate penalty functions, large-scale constraints are successfully incorporated into the optimization criterion, thus transforming the dynamic optimization problem into a static one with simple bounds on the decision variables. Simulation results verify that our proposed methodology can provide accurate results and can deal with various optimization objectives uniformly.

Simultaneous dynamic optimization

Our trajectory planner can tackle different requirements or constraints uniformly.Our proposal is systematically tested on a wide range of simulation cases.A Hamiltonian-based index is utilized to judge the optimality of an obtained solution.Differences between min-time and min-length trajectories are investigated. Trajectory planning in robotics refers to the process of finding a motion law that enables a robot to reach its terminal configuration, with some predefined requirements considered at the same time. This study focuses on planning the time-optimal trajectories for car-like robots. We formulate a dynamic optimization problem, where the kinematic principles are accurately described through differential equations and the constraints are strictly expressed using algebraic inequalities. The formulated dynamic optimization problem is then solved by an interior-point-method-based simultaneous approach. Compared with the prevailing methods in the field of trajectory planning, our proposed method can handle various user-specified requirements and different optimization objectives in a unified manner. Simulation results indicate that our proposal efficiently deals with different kinds of physical constraints, terminal conditions and collision-avoidance requirements that are imposed on the trajectory planning mission. Moreover, we utilize a Hamiltonian-based optimality index to evaluate how close an obtained solution is to being optimal.

Time-Optimal Maneuver Planning in Automatic Parallel Parking Using a Simultaneous Dynamic Optimization Approach

Autonomous parking has been a widely developed branch of intelligent transportation systems. In autonomous parking, maneuver planning is a crucial procedure that determines how intelligent the entire parking system is. This paper concerns planning time-optimal parallel parking maneuvers in a straightforward, accurate, and purely objective way. A unified dynamic optimization framework is established, which includes the vehicle kinematics, physical restrictions, collision-avoidance constraints, and an optimization objective. Interior-point method (IPM)-based simultaneous dynamic optimization methodology is adopted to solve the formulated dynamic optimization problem numerically. Given that near-feasible solutions have been widely acknowledged to ease optimizing nonlinear programs (NLPs), a critical region-based initialization strategy is proposed to facilitate the offline NLP-solving process, a lookup table-based strategy is proposed to guarantee the on-site planning performance, and a receding-horizon optimization framework is proposed for online maneuver planning. A series of parallel parking cases is tested, and simulation results demonstrate that our proposal is efficient even when the slot length is merely 10.19% larger than the car length. As a unified maneuver planner, our adopted IPM-based simultaneous dynamic optimization method can deal with any user-specified demand provided that it can be explicitly described.

Time-optimal trajectory planning for tractor-trailer vehicles via simultaneous dynamic optimization

Trajectory planning is a critical aspect of autonomous tractor-trailer vehicle design. Trajectory planning algorithms usually compute paths first, trajectories are obtained thereafter. This multi-step feature makes those planners inefficacious to handle time-dependent constraints. In this study, we consider the original trajectory planning mission directly, which is described as an optimal control problem containing the kinematics, mechanical/physical constraints, environmental requirements as well as an optimization criterion. In this formulation, only the fundamental driving principles with no special issues (e.g., backing-up maneuver and jackknife) are considered. For example, the prevailing “small-angle assumption” is not utilized to prevent jackknifing. Instead, we only require that different parts of a tractor-trailer vehicle should not collide, since the emergence of jackknife does not physically violate the kinematics. An interior-point method based simultaneous approach is adopted to solve the formulated optimal control problem. Simulation results verify our proposal is capable of handling scenarios with various user-specified requirements.

Spatio-temporal decomposition

Motion planning methodologies for parallel parking have been well developed in the last decade. In contrast to the prevailing and emerging parking motion planners, this work provides a precise and objective description of the parking scenario and vehicle kinematics/dynamics. This is achieved by formulating a unified optimal control problem that is free of subjective knowledge (e.g., human experiences). The concerned optimal control problem, when parameterized into a large-scale nonlinear programming (NLP) problem, is extremely difficult to solve. This bottleneck has hindered many research efforts previously. Although the feasible regions of NLP problems are clearly defined, the majority of NLP-solving processes still require high-quality initial guesses, which accelerate the convergence process. In this work, we propose a spatio-temporal decomposition based initialization strategy to generate reliable initial guesses, so as to facilitate the NLP-solving process. In contrast to the typical facilitation strategies in robotic motion/path planning, our spatio-temporal decomposition strategy considers only objective kowledge, further breaking the limitation of subjective knowledge and making full use of a vehicles maneuver potential. A series of comparative simulations verifies that the proposed initialization strategy is advantageous over its prevailing competitors, and that the proposed motion planner is promising for on-line planning missions. Theoretical analysis that supports our initialization strategy is given as well.

Centralized and optimal motion planning for large-scale AGV systems

A centralized motion planner is proposed for large-scale AGV systems;AGV kinematics and dynamics are clearly incorporated into the problem formulation;Formation reconfiguration benchmark cases are set up.Validation and unification of the proposed motion planner are investigated via exhaustive tests;On-line computation capability is promising with near-optimal initial guess. A centralized multi-AGV motion planning method is proposed. In contrast to the prevalent planners with decentralized (decoupled) formulations, a centralized planner contains no priority assignment, decoupling, or other specification strategies, thus is free from being case-dependent and deadlock-involved. Although centralized motion planning is computationally expensive, it deserves investigations in schemes that are sensitive to solution quality but insensitive to computation time. Specifically, centralized multi-AGV motion planning is formulated as an optimal control problem in this work, wherein differential algebraic equations are used to describe the AGV dynamics, mechanical restrictions, and exterior constraints. Orthogonal collocation direct transcription method is adopted to discretize the original infinite-dimensional optimal control problem into a large-scale nonlinear programming (NLP) problem, which is solved using interior point method thereafter. Exhaustive simulations are conducted on 10-AGV formation reconfiguration tasks. Simulation results show the validation, unification, and real-time implementation potential of the introduced centralized planner. Particularly, the computation time on a PC reduces to several seconds with near-optimal initial guess in the NLP solving process, making receding horizon replanning possible via this centralized planner.

Shape Matching Optimization via Atomic Potential Function and Artificial Bee Colony Algorithms with Various Search Strategies

Visual shape matching is a critical topic in pattern recognition applications. Atomic potential matching (APM) model is a relatively new shape matching methodology inspired by potential field attractions. Compared to the conventional edge potential function model, APM not only encourages the right matching parts through attraction, but also repels the wrong matching parts. This feature enables APM to cope with targets that hide in the intricate background. This study comprehensively investigates the convergence performances of various state-of-the-art artificial bee colony (ABC) algorithms in shape matching problems on the basis of APM framework. Repeated simulations are conducted to evaluate the optimization abilities of the concerned ABC variants and experimental results indicate that the prevailing remedies for the conventional ABC algorithm, especially efforts made in the local exploitation phase, are not efficacious to promote optimization capability. Explanations regarding the comparative results are provided as well.

Simultaneous versus joint computing: A case study of multi-vehicle parking motion planning

Abstract Multi-vehicle motion planning (MVMP) refers to computing feasible trajectories for multiple vehicles. MVMP problems are generally solved in two ways, namely simultaneous methods and joint methods. An inherent difference between both types of methods is that, simultaneous methods compute motions for vehicles all at once, while joint methods divide the original problem into parts and combine them together. The joint methods usually sacrifice solution quality for computational efficiency, and the simultaneous methods are applicable to simple or simplified scenarios only. These defects motivate us to develop an efficient simultaneous computation method which provides high-quality solutions in generic cases. Progressively constrained dynamic optimization (PCDO), an initialization-based computation framework is proposed to ease the burdens of simultaneous computation methodologies when they are adopted to solve the MVMP problems. Specifically, PCDO locates and discards the redundant constraints in the MVMP problem formulation so as to reduce the problem scale, thereby easing the problem-solving process. Our simulations focus on the cooperative parking scheme of automated vehicles. Comparative simulation results show that (1) the designs in PCDO are efficient, and (2) simultaneous computation outperforms joint computation.

Paving green passage for emergency vehicle in heavy traffic: Real-time motion planning under the connected and automated vehicles environment

This paper describes a real-time multi-vehicle motion planning (MVMP) algorithm for the emergency vehicle clearance task. To address the inherent limitations of human drivers in perception, communication, and cooperation, we require that the emergency vehicle and the surrounding normal vehicles are connected and automated vehicles (CAVs). The concerned MVMP task is to find cooperative trajectories such that the emergency vehicle can efficiently pass through the normal vehicles ahead. We use an optimal-control based formulation to describe the MVMP problem, which is centralized, straightforward, and complete. For the online solutions, the centralized MVMP formulation is converted into a multi-period and multi-stage version. Concretely, each period consists of two stages: the emergency vehicle and several normal CAVs ahead try to form a regularized platoon via acceleration or deceleration (stage 1); when a regularized platoon is formed, these vehicles act cooperatively to make way for the emergency vehicle until the emergency vehicle becomes the leader in this local platoon (stage 2). When one period finishes, the subsequent period begins immediately. This sequential process continues until the emergency vehicle finally passes through all the normal CAVs. The subproblem at stage 1 is extremely easy because nearly all the challenging nonlinearity gathers only in stage 2; typical solutions to the subproblem at stage 2 can be prepared offline, and then implemented online directly. Through this, our proposed MVMP algorithm avoids heavy online computations and thus runs in real time.