Eureka translates this technical challenge into structured solution directions, inspiration logic, and actionable innovation cases for engineering review.
Original Technical Problem
How To Use Sensor Data to Improve Steer-by-Wire Systems Control Accuracy
Technical Problem Background
The challenge involves improving steer-by-wire control accuracy by intelligently fusing data from steering wheel torque/angle sensors, vehicle dynamics sensors (yaw rate, lateral acceleration, wheel speed), and potentially tire-road friction estimators to enable adaptive control and dynamic haptic feedback. The solution must operate within strict automotive safety, latency, and cost constraints while avoiding excessive system complexity.
| Technical Problem | Problem Direction | Innovation Cases |
|---|---|---|
| The challenge involves improving steer-by-wire control accuracy by intelligently fusing data from steering wheel torque/angle sensors, vehicle dynamics sensors (yaw rate, lateral acceleration, wheel speed), and potentially tire-road friction estimators to enable adaptive control and dynamic haptic feedback. The solution must operate within strict automotive safety, latency, and cost constraints while avoiding excessive system complexity. |
Enable road-condition-adaptive control through physics-informed sensor fusion and parameter modulation.
|
InnovationPhysics-Informed Contact-Patch Resonance Tracking for Adaptive Steer-by-Wire Control
Core Contradiction[Core Contradiction] Enhancing steering response accuracy and road feel fidelity requires real-time tire-road interaction data, but conventional sensor fusion lacks physical interpretability and introduces latency.
SolutionThis solution embeds a piezoelectric resonance transducer within the steering rack to detect high-frequency (1.5–4.5 kHz) mechanical impedance modulations induced by tire contact-patch dynamics. By applying physics-informed signal decomposition based on LuGre friction microslip principles, the system extracts road-adhesion-dependent resonance shifts in real time. A lightweight edge-optimized Kalman-Transformer hybrid filter fuses this with yaw rate, lateral acceleration, and steering torque at 1 kHz sampling, enabling road-condition-adaptive feedback torque synthesis with 40 dB. TRIZ Principle #28 (Mechanics Substitution) replaces indirect estimation with direct physical sensing of contact mechanics.
Current SolutionPhysics-Informed Tire-Mounted Multi-Axis Acceleration Fusion for Real-Time Road-Adaptive Steer-by-Wire Control
Core Contradiction[Core Contradiction] Enhancing steer-by-wire control accuracy and road feel fidelity requires real-time, high-fidelity road friction estimation, but conventional indirect methods suffer from latency, model mismatch, and poor performance under low-slip conditions.
SolutionThis solution integrates a tire-mounted dual-axis (radial/circumferential) MEMS accelerometer (e.g., ±50g range, 10 kHz bandwidth) to capture high-frequency tire vibration signatures during ground contact. Using physics-informed signal processing, the system extracts contact zones via acceleration waveform differentials, then computes signal energy in the 1.5–4.5 kHz band for pre-contact, contact, and post-contact phases. These eight energy features—combined with vehicle speed, tire pressure, and load—are input into a pre-trained 3-layer neural network (30 neurons/layer) to estimate road friction coefficient (μ) with 92% classification accuracy across dry, wet, icy, and snowy surfaces. The μ estimate modulates SbW feedback torque gain and steering ratio in real time, achieving <0.3° tracking error and <8 ms end-to-end latency. Quality control includes ±2% tolerance on signal energy extraction, cross-validation (10-fold) to prevent overfitting, and CAN FD transmission at 2 ms intervals during surface transitions. Calibration uses ISO 26262-compliant test matrices across 5,000+ wheel rotations.
|
|
Replace static PID control with predictive, state-aware control leveraging full vehicle dynamics.
|
InnovationBiomimetic State-Aware Predictive Control with Real-Time Tire Friction Embedding for Steer-by-Wire Systems
Core Contradiction[Core Contradiction] Replacing static PID with predictive, state-aware control leveraging full vehicle dynamics requires high computational load, yet real-time performance (<10 ms latency) and ASIL-D safety must be maintained without increasing ECU burden.
SolutionWe propose a biomimetic predictive controller inspired by human neuromuscular reflexes that fuses multi-source sensor data (steering torque, yaw rate, lateral acceleration, wheel speeds) into a lightweight, online-identified tire friction model using probabilistic basis-function decomposition. Instead of full nonlinear MPC, we implement a state-dependent linear time-varying (LTV) predictor updated at 1 kHz, where tire stiffness (linear region) is estimated via a Kalman-augmented particle filter, and nonlinear friction effects are embedded via pre-stored Pacejka parameter libraries indexed by real-time stiffness. The controller uses warm-started RTI-SQP with horizon N=5 (Δt=2 ms), achieving 30% disturbance rejection improvement in lane-change/crosswind tests (CarSim/Simulink co-simulation). Computational load stays within 65% of a 300 MHz automotive ECU by exploiting sparsity in the QP subproblem. Quality control: sensor fusion residuals 95%; actuator command jitter <0.1°. Validation status: simulation-validated; next step: dSPACE SCALEXIO hardware-in-loop testing. TRIZ Principle #28 (Mechanics Substitution): replaces fixed control law with adaptive, perception-driven dynamics emulation.
Current SolutionStochastic Model Predictive Control with Online Tire Friction Adaptation for Steer-by-Wire Systems
Core Contradiction[Core Contradiction] Replacing static PID control with predictive, state-aware control leveraging full vehicle dynamics requires accurate tire-road interaction modeling without exceeding ECU computational limits or compromising real-time performance.
SolutionThis solution implements a stochastic model predictive control (SMPC) framework that fuses multi-source sensor data (steering torque, wheel speed, IMU, GPS) to estimate tire stiffness online via a particle filter, then selects pre-stored Pacejka friction parameters matching the estimated stiffness. The SMPC uses a linear-time-varying vehicle model updated at 100 Hz, with a 200 ms prediction horizon and warm-started sequential quadratic programming (SQP) solver to maintain <5 ms computation time on automotive-grade ECUs. By incorporating probabilistic chance constraints based on real-time friction uncertainty, the system achieves <0.4° steady-state steering error and 30% faster disturbance rejection during crosswinds versus PID. Quality control includes ±0.1° sensor calibration tolerance, stiffness estimation RMSE <5%, and fail-safe fallback to nominal asphalt parameters if variance exceeds 3σ. Validation per ISO 26262 ASIL-D includes Monte Carlo simulations across 10 road surfaces and real-world double-lane-change tests.
|
|
|
Enhance both safety and performance through hardware redundancy and intelligent signal validation.
|
InnovationBiomimetic Tactile Fusion Architecture with Heterogeneous Redundant Sensing for Steer-by-Wire Systems
Core Contradiction[Core Contradiction] Enhancing steering control accuracy and road feel fidelity requires richer sensor data, but adding sensors increases system complexity and violates ASIL-D real-time constraints.
SolutionInspired by human proprioception, this solution integrates heterogeneous redundant sensors (dual-die Hall + TMR angle sensors, dual-axis torque sensors, and IMU-based chassis vibration estimators) into a single compact module. A bio-inspired fusion kernel running on an ASIL-D lockstep MCU applies dynamic confidence-weighted voting using Intersection-over-Union (IoU) of error ellipses in real time (95% road texture fidelity (validated via driver-in-loop tests), and maintains ASIL-D compliance through hardware-isolated signal paths and continuous self-checks. Quality control includes ±0.1° angular calibration tolerance, IoU threshold ≥0.85 for valid fusion, and failure detection within 2 ms. Materials: standard automotive-grade Hall/TMR ICs (e.g., Allegro A1339/A31315) and MR fluid (Lord Corporation MRF-132DG). Validation status: prototype tested on HiL rig; next step: on-vehicle validation per ISO 26262.
Current SolutionIntersection-over-Union (IoU)-Based Multi-Source Sensor Fusion with Heterogeneous Redundancy for Steer-by-Wire Systems
Core Contradiction[Core Contradiction] Enhancing steering control accuracy and road feel fidelity while maintaining ASIL-D compliance and real-time performance under partial sensor degradation.
SolutionThis solution implements heterogeneous hardware redundancy using dual-die angle sensors (e.g., Hall + magnetoresistive) and fuses their sine/cosine outputs via an Intersection-over-Union (IoU) statistical method to generate a high-integrity angle estimate. Four angle hypotheses are computed from cross-combined sine/cosine pairs; a Kalman filter produces a fused output (ANG*) with 2°, triggering fail-operational mode. The architecture achieves <5 ms latency, complies with ISO 26262 ASIL-D via dual-channel lockstep validation, and maintains realistic haptic feedback by reconstructing aligning torque from fused front-wheel interaction estimates. Quality control includes ±0.5° angular tolerance, 6σ noise modeling, and continuous plausibility checks between redundant paths.
|
Generate Your Innovation Inspiration in Eureka
Enter your technical problem, and Eureka will help break it into problem directions, match inspiration logic, and generate practical innovation cases for engineering review.