UAV-mounted Mobile Edge Computing (MEC) can extend vehicular computing where terrestrial infrastructure is sparse, but dense traffic and limited UAV batteries make resource management difficult. We consider a swarm in which a base-station-connected Head-UAV coexists with a flying ad hoc network of Member-UAVs and propose Adaptive Mode-Aware Swarm Hybrid (A-MASH), which jointly decides vehicle association with admission control, MEC CPU allocation, uplink power and UAV-to-UAV relaying. A battery- and load-aware ant colony search over the discrete decisions is coupled with a Grey Wolf Optimizer over the continuous resources; the Head-UAV access mode follows load, battery and interference-cancellation quality, and a Kalman-filter channel predictor supports decoding. In slot-level simulations with 10–150 vehicles and equal evaluation budgets, A-MASH delivers 65–102% more tasks on time than standalone metaheuristics at 100 vehicles and matches an alternating-optimization baseline up to 100 vehicles, but simple model-based heuristics are better once the load exceeds the computing capacity of the swarm. The predictor adds up to 5.5 percentage points of delivery ratio, NOMA helps only with near-perfect interference cancellation, and the battery-aware mechanisms postpone the first UAV withdrawal by one scheduling epoch.
Regenerative-braking energy is determined by battery-pack current, voltage, and event duration, yet temporal networks have been proposed to estimate it from compressed event sequences. We audited four archived architectures (1D CNN, LSTM, GRU, and TCN) on 8181 events from one publicly available electric-vehicle dataset and reanalyzed their validation design. Events were represented by 32 samples of current, voltage, state of charge, temperature, and elapsed time. The original three-repeat comparison changed the grouped holdout between repeats and used globally derived energy-class thresholds; one extreme event moved between development and test. A direct 32-point trapezoidal-integration baseline, calculated from the identical network input, achieved R2 = 0.999879, RMSE = 0.554 Wh, and MAE = 0.147 Wh over all events. On fixed, non-overlapping recording-group and strictly forward whole-group tests, its RMSEs were 0.302 and 0.414 Wh, respectively. We additionally evaluate paired four- versus five-channel neural models, group-level dependence, distribution shift, and energy-magnitude versus peak-power classifications. The results show that this full-event, post-event task is largely an electrical-integration problem; high energy-derived classification scores should not be interpreted as recognition of an independent braking state. These single-vehicle, offline results do not establish cross-platform transfer or real-time control feasibility.
Following a trajectory and avoiding obstacles are two closely linked control tasks for an autonomous vehicle. Classical reactive controllers are tuned for one operating range, do not look ahead and have no built-in way to keep a minimum distance from an obstacle. This paper presents a Linear Time-Invariant Model Predictive Control (LTI-MPC) scheme combined with a Luenberger observer for the kinematic bicycle model. The model is linearised once around (0,0,0,vnom) and written in deviation coordinates so that the drift term of the predictor is exact. The observer, tuned with the discrete-time algebraic Riccati equation, rebuilds the heading from the noisy DGPS-RTK position and odometry speed, so no heading sensor is needed. Static obstacles are handled by a local re-planner with a designed clearance, and a closed-form rule links its blending length to the maximum reference heading. Tested in a closed loop with observer and measurement noise on five scenarios, the LTI-MPC reaches a tracking RMSE of 0.018 m at 4 m/s on a sinusoidal path; achieves, in simulation, a clearance of 0.98–0.99 m from the obstacle (the reference is designed for 1.0 m, the requirement is 0.5 m); and is the only controller among PID-type, Stanley and MPC whose accuracy degrades smoothly from 2 to 8 m/s with a single tuning. The observer converges within 1∘ in 0.24 s. A sweep of constant-radius turns shows that the linearisation stays accurate at up to 30∘ of heading change.
During vehicle operation, lateral tire forces generally require a finite time to reach their steady-state values, and this dynamic response can be characterized by the tire relaxation length. For an active front steering (AFS) vehicle equipped with a mechanical elastic wheel (MEW), an MPC-based control strategy is developed to account for this delayed tire response and improve handling stability. The front steering angle is selected as the control input, while a first-order lag formulation is incorporated into the prediction model to represent the transient evolution of lateral tire forces. To accommodate variations in tire response, the lateral relaxation length is estimated online using an unscented Kalman filter (UKF). The lateral tire forces are calculated using the Magic Formula tire model, which describes the nonlinear relationship between tire forces and slip conditions. To reduce the computational burden associated with online optimization, the tire model is linearized around the current operating point before being used in the controller. To examine the robustness of the UKF estimator under vehicle parameter uncertainty, a 20% increase in vehicle mass is applied. The proposed control method is then assessed in three simulation scenarios, with a conventional MPC controller and an LQR controller serving as benchmarks. Although the vehicle mass is perturbed, the estimated lateral relaxation lengths of the front and rear axles follow broadly similar variation patterns. The simulation results also show that the proposed method achieves improved yaw rate tracking and sideslip angle regulation compared with conventional MPC, thereby enhancing vehicle handling stability.
Ethical decision-making and responsibility assessment in autonomous driving are often addressed separately, which limits the consistency between maneuver selection and post-event analysis. This study proposes a responsibility-aware decision-making framework that integrates a multi-actor responsibility model with a comprehensive ethical-cost function. Moral-risk scenarios are classified and represented using variables related to the vehicle state, road environment, traffic participants, human–machine interactions, and data traceability. Seven responsibility factors are weighted using expert elicitation and the analytic hierarchy process. Responsibility shares for five actor groups are then calculated from the weighted contribution matrix. Candidate maneuvers are first screened by vehicle-dynamics constraints and then evaluated using five normalized cost components: collision risk, harm to life, rule violation, vehicle-stability risk, and responsibility transfer. The framework is evaluated in a pedestrian red-light crossing scenario through SCANeR–MATLAB co-simulation. Under the baseline condition, emergency braking produced the lowest comprehensive ethical cost of 0.276, while the external traffic participant and autonomous-driving system received relative responsibility weights of 0.50 and 0.26, respectively. Across five operating conditions, the preferred maneuver varied between emergency braking and deceleration while maintaining the lane as the vehicle speed and surrounding traffic gaps changed. The results indicate that the framework can adapt maneuver selection to scenario-dependent risk while maintaining a traceable assessment of relative responsibility contributions. The framework provides a quantitative engineering approach for integrating ethical maneuver selection with responsibility assessment; the resulting responsibility weights are scenario-dependent analytical quantities and are not intended to determine legal liability. These results may inform the design and simulation-based evaluation of responsibility-aware autonomous-driving strategies.
