Autonomous Vehicle Systems

Autonomous vehicle systems integrate sensing, perception, decision-making and control to deliver self-driving capabilities across diverse environments. Core components include multi-modal sensor suites—cameras, LiDAR, radar and inertial units—that generate a rich situational picture. Perception algorithms detect and classify road users and static features, while simultaneous localisation and mapping (SLAM) and vehicle-to-vehicle/infrastructure communications maintain accurate positioning. Decision layers assess high-level manoeuvres—lane changes, turns or obstacle avoidance—and translate them into optimized trajectories. These trajectories are executed by model predictive and adaptive control schemes that respect vehicle dynamics, actuator limits and safety envelopes. Hierarchical architectures combine global route planning with local real-time replanning to handle dynamic hazards. Collectively, these advances promise enhanced road safety, reduced energy consumption and scalable deployment of autonomous functions—from advanced driver assistance to fully driverless operations—in urban, rural and highway settings.

Research from Nature Portfolio

A new integrated collision-avoidance strategy combines adaptive model predictive control with four-wheel steering and differential braking to enhance path-tracking accuracy and stability under emergency scenarios. Extensive MATLAB/Simulink and CarSim co-simulations demonstrate reduced tracking error, yaw-rate overshoot and roll excursions compared with conventional approaches.

A dynamic-boundary-based lateral motion synergistic control system categorises distributed-drive vehicle states into stable, quasi-stable and unstable domains. A coordination layer modulates the strength of path-following and yaw-stability commands according to real-time adhesion estimates, yielding improved accuracy and stability in double-lane-change tests.

An eight-degrees-of-freedom multidirectional motion coupling control framework integrates path-tracking, speed control, yaw stabilization and active suspension in a single system. Extreme speed estimation via dynamic boundaries and multi-DOF vehicle modelling delivers robust performance in serpentine and double-lane-shift simulations under variable adhesion.

Research from all publishers

A comprehensive review of model predictive control (MPC) for automated road vehicles highlights systematic handling of multi-variable interactions, real-time constraint satisfaction and horizon tuning. This survey underscores MPC’s leading role in lane-keeping and trajectory tracking amid stringent road-geometry and actuator limits.

An event-triggered unscented Kalman filter, combined with an interacting multiple-model approach, leverages intermittent vehicle-to-vehicle data exchanges to sustain high-fidelity state estimation with minimal bandwidth. Simulations show preserved accuracy even when communication rates drop below 15% of nominal levels.

Advances in wheel-slip control for anti-lock braking systems employ nonlinear model predictive control with load-sensing technology to dynamically adjust slip targets. Comparative tests against industrial benchmarks reveal shorter stopping distances and enhanced directional control across variable surface conditions.

Autonomous Vehicle Systems publication trend

The graph below shows the total number of articles in autonomous vehicle systems across all publications each year (not limited to Nature Index journals).

Technical terms

Model predictive control (MPC): A receding-horizon strategy that solves an optimisation problem at each step to predict and enforce future states while honouring system constraints.

Adaptive model predictive control (AMPC): An MPC variant that updates model parameters online to maintain performance under changing dynamics or uncertainties.

Unscented Kalman filter (UKF): A nonlinear estimator that propagates a set of sigma points through the state-transition model to approximate posterior means and covariances.

Event-triggered control: A scheme that schedules state-estimation or control updates only when predefined error or condition thresholds are exceeded, reducing communication load.

Dynamic boundary: A real-time threshold surface, derived from vehicle states and adhesion parameters, used to distinguish stability domains and coordinate control actions.

Differential braking: The selective application of braking torque to individual wheels, used to generate steering moments or assist in evasive manoeuvres.

References

1. Interacting multiple model-based ETUKF for efficient state estimation of connected vehicles with V2V communication. Green Energy and Intelligent Transportation (2023).
2. Model predictive path tracking control for automated road vehicles: A review. Annual Reviews in Control (2023).
3. Survey on Wheel Slip Control Design Strategies, Evaluation and Application to Antilock Braking Systems. IEEE Access (2020).
4. Emergency collision avoidance strategy for autonomous vehicles based on steering and differential braking. Scientific Reports (2022).
5. Dynamic-boundary-based lateral motion synergistic control of distributed drive autonomous vehicle. Scientific Reports (2021).
6. Multidirectional motion coupling based extreme motion control of distributed drive autonomous vehicle. Scientific Reports (2022).