This paper investigates the design of stealthy deception attacks with the aim of destroying the estimate performance without knowing the filter gain. The residual used to detect attacks is generated by the parity space approach and the Kullback-Leibler divergence is adopted as the metric of stealthiness. We first give the necessary and sufficient condition for the inexistence of the strictly stealthy attack, which does not change the residual and can lead to unbounded estimate error. When the strictly stealthy attack does not exist, a lower bound of the secondary moment of the estimate error is then derived via the posterior Cramer-Rao bound. The zero-mean Gaussian attack that maximizes this lower bound is obtained by solving a convex optimization problem. The proposed method can also be applied to design stealthy attacks with the aim of destroying the control performance. Finally, a numerical example of longitudinal flight control system is illustrated to demonstrate the effectiveness of the proposed attack.

This article studies the design of the optimal stealthy joint attacks against distributed state estimation in cyber-physical systems to steer the state estimates in all the sensors as close as possible to a common target within finite time. To ensure stealthiness, both the Kullback-Leibler divergence between the innovations and the Kullback-Leibler divergence between the consensus errors with and without attacks at each instant should not exceed given thresholds, respectively. In the case that all the thresholds equal to zero, the optimal attack is given in the form of an analytical expression. In the cases that some thresholds are larger than zero, an optimization problem is solved for each case at each instant, and the corresponding attack sequence obtained is proved to be globally optimal. Finally, a numerical example is adopted to verify the effectiveness of the proposed optimal attack strategies.

Accurately evaluating the intelligence of autonomous path planning remains challenging, primarily due to the interdependencies among evaluation metrics and the insufficient integration of subjective and objective weighting methods. This paper proposes Game-Theoretic Kendall’s Coefficient (GTKC) weighting framework for evaluating autonomous path planning intelligence. The framework specifies a safety–efficiency–comfort metric system with observable, reproducible, and quantifiable metrics. To account for intermetric dependence, subjective weights are elicited via an improved Analytic Network Process (ANP), while objective weights are derived using the CRITIC method to capture contrast intensity and intercriteria conflict. The credibility of the subjective and objective weights is evaluated using Kendall’s coefficient and the coefficient of variation, respectively. Subsequently, based on the principle that higher credibility should receive greater weight, a game-theoretic optimization model is employed to dynamically derive optimal combination coefficients. Experimental results on three case scenarios demonstrate that the GTKC framework significantly outperforms existing weighting approaches in terms of effectiveness (achieving a lowest Mean Absolute Error (MAE) of 0.15 and a perfect Spearman’s correlation coefficient ($\overline{\rho }=1.0$) with ground-truth rankings), stability (Mean Standard Deviation (MSD) = 0.023), and ranking consistency (Kendall’s coefficient W = 0.924). These findings validate GTKC as a theoretically grounded and practically robust mechanism that explicitly models metric interdependencies and integrates expert knowledge with empirical evidence, enabling reliable and reproducible evaluation of autonomous path planning intelligence.

Autonomous vehicles remain commercially limited largely due to safety performance stagnation. Existing deep learning, heavily reliant on failure data from rare safety-critical events, suffers from the seesaw effect—improvement in some scenarios causes regression in others. We introduce an innovative dense learning approach that prioritizes both informative failures and successes, informed by theoretical findings. Data is sampled proportionally to its contribution to the policy gradient and exposure frequency, excluding non-informative samples. This densifies the training dataset’s information, significantly reducing learning variance without bias, enabling tasks intractable for existing methods. To validate this, we trained a safety-critical driving agent for a highly automated vehicle using mixed reality on an urban test track. Results demonstrate that our approach breaks the performance stagnation, enhancing the model’s overall safety performance by one to two orders of magnitude. This marks a significant stride towards achieving human-level safety and widespread adoption for autonomous vehicles.

Aiming to address the core issue that the current intelligence evaluation for Unmanned Ground Vehicles (UGVs) overly rely on static performance metrics and lack dynamic quantitative characterization of learning internalization capability (LIC), this study proposes a dynamic evaluation framework based on time series analysis. The framework begins by constructing a multidimensional test scenario parameter system and collecting externally observable performance sequence data. It then introduces a sliding window-based slope-standard deviation collaborative analysis technique to achieve unsupervised division of learning phases, from which five core evaluation metrics are extracted to comprehensively quantify the multidimensional dynamic characteristics of LIC in terms of efficiency, stability, and overall effectiveness. Simulation experiments were carried out using UGVs equipped with three types of path-planning algorithms in low-, medium-, and high-difficulty scenarios. Results demonstrate that the proposed algorithm can effectively distinguish multi-dimensional differences in LIC among different UGVs, exhibiting strong discriminative power and interpretability. This study provides a standardized evaluation tool for UGV intelligent selection, algorithm iteration optimization, and training strategy design, and offering significant reference value for the evaluation of the learnability of autonomous driving systems.

This paper investigates optimal stealthy estimator attacks against interconnected systems. An attacker injects false data into the state estimation process to maximize a finite-horizon quadratic utility involving system states and control inputs, while penalizing the attack energy. Stealthiness is enforced by two detectors based on residuals and distributed interconnection estimation errors. Exploiting the decoupled evolution of local estimation errors, we reformulate the attack design as a constrained quadratic optimization over the attack sequence with multi-step stealthiness constraints over the remaining horizon with given thresholds. We further establish recursive feasibility of the proposed scheme and prove that the resulting receding-horizon policy is globally optimal for the original finite-horizon problem under the adopted information pattern. Simulations on a benchmark interconnected system validate effectiveness of the proposed attack policy and illustrate the impact-stealthiness tradeoff.