This paper presents a model-based algorithm that estimates how the driver of vehicle can either steer, brake, or accelerate to avoid colliding with an arbitrary object. In this algorithm, motion is described by linear bicycle model, and perimeter represented rectangle. The estimated object polygon allowed change size, shape, position, orientation at sampled time instances. Potential evasive maneuvers are modeled, parameterized, approximated such analytical expression be derived estimate set...

In this article, a scenario where several vehicles have to coordinate among them in order cross traffic intersection is considered. case, the control problem relies on optimization of cost function while guaranteeing collision avoidance and satisfaction local constraints. A decentralized solution proposed sequentially solve problems allowing cross, safe way, intersection. This approach pays special attention how degrees freedom that each vehicle disposes avoid potential can be quantified led...

Todays cruise control systems try to keep a constant speed given by the driver without regarding energy consumption. There is, however, possibilities save choosing optimal velocity is neglected. Solving underlying problem for long distances with an algorithm that runs on vehicle unit not straightforward, especially when it comes hybrid electric vehicles (HEV). In order overcome computational burden, this paper presents two-level model predictive approach. It uses sequential quadratic program...

This article focuses on the traffic coordination problem at intersections. We present a decentralized approach, combining optimal control with model-based heuristics. show how heuristics can lead to low-complexity solutions that are suitable for fast online implementation, and analyze its properties in terms of efficiency, feasibility optimality. Finally, simulation results different scenarios also presented.

The Levenberg-Marquardt algorithm is often superior to other training algorithms in off-line applications. This motivates the proposal of using a recursive version for on-line neural nets nonlinear adaptive filtering. performance suggested compared with alternative algorithms, such as steepest-descent and Gauss-Newton algorithms. advantages disadvantages different are pointed out. tested on some examples, it shown that generally has better convergence properties than

In this paper we discuss the role of criterion minimization as a means for parameter estimation. Most traditional methods, such maximum likelihood and prediction error identification are based on these principles. However, somewhat surprisingly, it turns out that is not always 'optimal' to try find absolute minimum point criterion. The reason 'stopped minimization' (where iterations have been terminated before has reached) more or less identical properties using regularization (adding...

We propose two model-based threat assessment methods for semi-autonomous vehicles, i.e., human-driven vehicles with autonomous driving capabilities. Based on information about the surrounding environment, we introduce a set of constraints vehicle states, which are satisfied under “safe” conditions. Then, formulate problem as constraint satisfaction problem. Vehicle and driver mathematical models used to predict future violation, indicating possibility accident or loss control, hence, need...

In this article, we consider the problem of coordinating a number vehicles crossing traffic intersection. The proposed solution is based on receding horizon formulation with pre-defined decision order. approach, local problems are formulated for each vehicle, which divided into finite-time optimal control problem, where collision avoidance enforced as terminal constraints, and an infinite can be solved offline. Feasibility conditions given sequence also derived simulation results presented.

This paper presents an algorithm for strategic decision making regarding when lane change and overtake manoeuvres are desirable feasible. By considering the task of driving on two-lane, one-way roads, as selection desired velocity profile, provides useful results in terms control well a variable corresponding to whether manoeuvre should be performed. The process is modelled through mixed logical dynamical system which solved model predictive using integer program formulation. performance...

State-of-health estimates of batteries are essential for onboard electric vehicles in order to provide safe, reliable, and cost-effective battery operation. This paper suggests a method estimate the 10-s discharge resistance, which is an established figure merit from laboratory testing, without performing test. Instead, state-of-health obtained using data directly their operational use, e.g., vehicles. It shown that simple dynamical models, based on current input voltage output, with model...

This paper develops a method for safe and autonomous intersection crossing. A centralized system controls vehicles within certain surrounding of the generates optimized trajectories all in area. recently proposed design approach, [10], where this problem is expressed as convex optimization using space sampling instead time sampling, formulated MPC solved by QP algorithms so that it can be executed real time. The controller then integrated CarMaker Matlab/Simulink validated against advanced...

This paper presents a novel heuristic method for optimal control of mixed-integer problems that, given feasible values the integer variables, are convex in rest variables. The is based on Pontryagin's maximum principle and allows problem to be solved using optimization techniques. advantage this approach short computation time obtaining solution near global optimum, which may otherwise need very long when by algorithms guaranteeing such as dynamic programming (DP). In paper, applied battery...

This paper studies the problem of optimally controlling an autonomous vehicle, to safely overtake a slow-moving leading vehicle. The is formulated minimize deviation from reference velocity and position trajectory, while keeping vehicle on road avoiding collision with surrounding vehicles. We show that optimization can be as convex program, by providing modeling steps include change frame, variables, sampling in relative longitudinal distance, relaxation linearization. A case study provided...

We stud the problem of optimally controlling autonomous vehicles to safely cross an intersection. The is approached by solving optimal control subproblem for all permutations crossing sequences. For a chosen sequence, we show that longitudinal vehicle control, subject collision avoidance constraints, can be formulated as convex program. proposed method transforms from original time domain space domain, and introduces change optimization variables replacing vehicles' speed with its inverse. A...

When controlling the energy flow among multiple power sources in electrified powertrains, typically non-convex set of original formulation is relaxed to a convex super-set, and problem then approached by means optimization. In this paper we show that when using backward simulation approach, where vehicle velocity equal reference velocity, global optimum can be obtained solving problem. kept as state problem, so called forward provide condition for which, satisfied, solution will non-relaxed

This paper concerns automated vehicles negotiating with other vehicles, typically human driven, in crossings the goal to find a decision algorithm by learning typical behaviors of vehicles. The vehicle observes distance and speed on intersecting road use policy that adapts its along pre-defined trajectory pass crossing efficiently. Deep Q-learning is used simulated traffic different predefined driver intentions. results show able cross intersection avoiding collision 98% time, while at same...