Real-Time Kinematics-Based Sensor-Fault Detection for Autonomous Vehicles Using Single and Double Transport with Adaptive Numerical Differentiation

TL;DR

This work presents KSFD, a real-time, model-independent sensor-fault-detection framework for autonomous vehicles that relies on exact kinematic relations and adaptive numerical differentiation. By using the single- and double-transport theorems together with AISE-based real-time differentiation, KSFD computes a set of kinematics-based error metrics from onboard sensors to uniquely identify a single faulty sensor in both ground and aerial platforms. The approach avoids model-based observers, reducing design complexity while maintaining robustness through adaptive noise estimation and a residual-based fault-diagnosis scheme. Simulated and experimental results on ground and aerial platforms demonstrate KSFD’s ability to promptly detect and isolate single-sensor faults, highlighting its potential for improving safety and reliability in autonomous systems.

Abstract

Sensor-fault detection is crucial for the safe operation of autonomous vehicles. This paper introduces a novel kinematics-based approach for detecting and identifying faulty sensors, which is model-independent, rule-free, and applicable to ground and aerial vehicles. This method, called kinematics-based sensor fault detection (KSFD), relies on kinematic relations, sensor measurements, and real-time single and double numerical differentiation. Using onboard data from radar, rate gyros, magnetometers, and accelerometers, KSFD uniquely identifies a single faulty sensor in real time. To achieve this, adaptive input and state estimation (AISE) is used for real-time single and double numerical differentiation of the sensor data, and the single and double transport theorems are used to evaluate the consistency of data. Unlike model-based and knowledge-based methods, KSFD relies solely on sensor signals, kinematic relations, and AISE for real-time numerical differentiation. For ground vehicles, KSFD requires six kinematics-based error metrics, whereas, for aerial vehicles, nine error metrics are used. Simulated and experimental examples are provided to evaluate the effectiveness of KSFD.

Figures (29)

• Figure 1: Coordinate frames for the vehicle kinematics under the assumption that the vehicle remains horizontal. $\hat{\imath}_{\rm E}$ points north, while $\hat{\jmath}_{\rm E}$ points east. The vectors $\hat{k}_{{\rm E}}$ and $\hat{k}_{{\rm B}}$ point vertically downward. $r_{x}, r_{y},$ and $\Psi$ are radar and magnetometer measurements, respectively, as summarized in Table \ref{['Tab:SensorData_ground']}.
• Figure 2: Block diagram of AISE.
• Figure 3: Position trajectory of the vehicle. The red $\times$ has zero inertial acceleration and serves as the radar target. The blue curve represents the simulated trajectory.
• Figure 4: Example \ref{['eg_sim_accelerometer_failure']}: Accelerometer with drift. (a) Beginning at 30 s, $L_{{\rm d},{x},k}$ exhibits drift. (b) $R_{{\rm d},{y},k}$ follows $L_{{\rm d},{y},k}$.
• Figure 5: Example \ref{['eg_sim_accelerometer_failure']}: Accelerometer with drift. $e_{{\rm d},{x},k}$ in (a) gradually increases at 30 s when the drift begins and exceeds the cutoff threshold $c_{{\rm d},{x}}$, which indicates a fault in the sensors used to compute $e_{{\rm d},{x},k}$. (b) $e_{{\rm d},{y},k}$ remains below cutoff.

Theorems & Definitions (5)

• Proposition 4.1
• Example 6.1
• Example 7.1
• Example 7.2
• Example 8.1