Risk-averse Stochastic Nonlinear Model Predictive Control for Real-time Safety-critical Systems

Abstract

Stochastic nonlinear model predictive control has been developed to systematically find an optimal decision with the aim of performance improvement in dynamical systems that involve uncertainties. However, most of the current methods are risk-neutral for safety-critical systems and depend on computationally expensive algorithms. This paper investigates on the risk-averse optimal stochastic nonlinear control subject to real-time safety-critical systems. In order to achieve a computationally tractable design and integrate knowledge about the uncertainties, bounded trajectories generated to quantify the uncertainties. The proposed controller considers these scenarios in a risk-sensitive manner. A certainty equivalent nonlinear model predictive control based on minimum principle is reformulated to optimise nominal cost and expected value of future recourse actions. The capability of proposed method in terms of states regulations, constraints fulfilment, and real-time implementation is demonstrated for a semi-autonomous ecological advanced driver assistance system specified for battery electric vehicles. This system plans for a safe and energy-efficient cruising velocity profile autonomously.

Ecological Advanced Driver Assistance System for Optimal Energy Management in Electric Vehicles

“…Another example of a SMPC for the ACC and CACC systems under uncertainty based on the constant time gap policy were introduced in [24]. A real-time Stochastic Nonlinear Model Predictive Control (SNMPC) with probabilistic constraints and Risksensitive Nonlinear Model Predictive Control (RSNMPC) were presented in [25] and [26] to compute a safe and energyefficient cruising velocity profile online.…”

“…Although the proposed methods mentioned in works of literature are effective for near-term prediction, rapid divergence can be experienced in far-term future prediction. A physicalstatistical motion model of the preceding vehicle robust to far-term future prediction was developed in [25] and [26]. The proposed model is based on the 85 th percentile speed concept and road geometry information.…”

“…[29]. To conclude, the introduced model is continuous and differentiable that is capable of propagating a plausible trajectory for the preceding vehicle motion along the prediction horizon (for more details, see [25] and [26]).…”

Stochastic optimal control for autonomous driving applications via polynomial chaos expansions

“…For these cases, Safaoui et al 14 proposed a rapidly exploring random tree algorithm enhanced with a risk measure constraint and the unscented transform (UT) for uncertainty propagation to facilitate stochastic motion planning of planar robots. In Reference 15 the authors employ a risk‐sensitive log‐expectation of the exponentiated performance index and a heuristic physical‐statistical model for uncertainty modeling to compute risk‐averse speed profiles for an adaptive cruise control (ACC) system.…”

Autonomous Navigation of AGVs in Unknown Cluttered Environments: Log-MPPI Control Strategy

“…On the other hand, despite the superior ability of SMPC to leverage the probabilistic nature of uncertainties, many SMPC approaches have performance limitations, including being tailored to specific forms of stochastic noise, requiring dynamics linearization to ensure a real-time performance and implementation of the control strategy, and employing chance constraints that may not accurately reflect the severity of constraint violations or potential accidents and can be computationally demanding to evaluate, particularly for complex or high-dimensional probability distributions [4], [5]. Additionally, SMPC has a further limitation in that it relies on a risk-neutral expectation to predict future uncertain outcomes, which may not be reliable in the case of a tail-end probability event (i.e., a rare event with a low probability of occurring) actually happening [6]. To address the challenges posed by uncertainties, Risk-Sensitive MPC (RSMPC) approaches have gained traction in recent years, thanks to their ability to balance the benefits and drawbacks of robust and stochastic MPC methods.…”