As a chassis design engineer specializing in commercial vehicles, I have encountered user feedback indicating that during autonomous driving, the electric power steering (EPS) system occasionally exits automatically, ceasing to accept external control signals and leading to a loss of vehicle control. For autonomous vehicles, the EPS is a core system that ensures the vehicle can follow external commands for directional changes. To guarantee stable vehicle control, this article delves into the control structure, working principles, and logic of the EPS, proposing an optimized control logic scheme. The goal is to ensure that the EPS maintains efficient and stable operation during normal driving and in scenarios where hydraulic assistance fails. Throughout this analysis, the role of the motor control unit is critical, as it governs the electric motor’s intervention and overall system responsiveness.
The EPS in autonomous vehicles must reliably execute external control instructions without unwarranted automatic exits. Additionally, it must ensure that in the event of hydraulic system failure, the electric motor promptly intervenes to provide assistance, allowing the driver to swiftly take over control. This dual requirement underscores the importance of a robust motor control unit that can seamlessly transition between autonomous and manual modes.
To understand the automatic exit issue, we must first examine the control logic of the EPS. The control strategy incorporates a motor torque threshold judgment to meet steering time and torque control requirements during hydraulic failure. The motor torque threshold is defined as 0.7 N·m. During operation, if the motor torque reaches this threshold, the system assumes hydraulic failure and switches to a rated torque output mode, ensuring the driver can regain control. When the motor torque is within the threshold range (typically less than 0.5 N·m), it provides normal assistance, amplified by a worm-gear reduction mechanism to act on the steering gear input shaft, ultimately delivering torque through the steering gear’s screw and sector mechanism.
However, when the motor torque exceeds the threshold, it outputs a rated torque of 8 N·m, again amplified by the reduction mechanism. In this scenario, the EPS exits the external control state, which is problematic for autonomous driving. To determine if the actual torque exceeds the threshold, we need to calculate the required torque for overcoming ground resistance and system inertia.
The torque required to overcome ground resistance during steering is derived from empirical formulas. For a vehicle’s front axle, the static steering resistance moment is given by:
$$M_r = \frac{3u}{3r} \frac{G}{P}$$
where \(u\) is the friction coefficient between tire and ground (typically 0.7), \(G\) is the front axle load, and \(P\) is tire pressure. The dynamic steering resistance moment during driving is:
$$M_R = R \times M_r$$
with \(R\) as an empirical correction coefficient (approximately 0.66 from extensive vehicle tests). Next, we calculate the linkage ratios of the front axle mechanism. The right wheel resistance transfers to the left steering knuckle via the trapezoidal mechanism, with the tie rod as a two-force member. The force transmission ratio \(i_T\) and the resistance moment \(M_c\) on the steering knuckle are:
$$i_T = \frac{L_{\text{right}} \csc D}{L_{\text{left}} \csc C}, \quad M_c = M_s \left(1 + \frac{1}{i_T}\right)$$
where \(L_{\text{left}}\) and \(L_{\text{right}}\) are the left and right lower arm force arm lengths, \(\csc C\) and \(\csc D\) are cosecant functions of angles, and \(M_s\) is the steering resistance moment of a single front wheel. The steering mechanism linkage ratio \(i_D\) and the total output assistance torque \(M_H\) from the steering gear are:
$$i_D = \frac{L_1 \csc B}{L_2 \csc A}, \quad M_H = M_c \frac{i_T + 1}{i_D i_T}$$
where \(L_1\) and \(L_2\) are force arm lengths, and \(\csc A\) and \(\csc B\) are cosecant functions. The total transmission ratio \(i_P\) is:
$$i_P = \frac{M_R}{M_H} = \frac{i_D i_T}{i_T + 1}$$
Considering transmission efficiencies \(\eta_1\) for the drag link (0.8) and \(\eta_2\) for the trapezoidal mechanism (0.8), the required steering gear assistance torque is:
$$M_H = \frac{M_s (i_T + 1)}{\eta_1 \eta_2 i_D i_T}$$
When the steering pump stops, the required torque is provided by the driver, and the steering wheel force is calculated as:
$$F = \frac{M_H + M_d}{i_p i_\omega R}$$
where \(i_p\) is the steering linkage ratio, \(i_\omega\) is the steering gear angle ratio, \(M_d\) is resistance from components and fluid, and \(R\) is the steering wheel radius. Based on these calculations, the required steering gear torque during normal driving is \(T_{\text{max}} = 8.5 \, \text{N·m}\), corresponding to a motor assistance torque of 0.425 N·m, which is below the 0.7 N·m threshold. However, this calculation omits inertial torques from rotating and moving parts, which can push the actual torque over the threshold.
The EPS involves multiple rotating and moving components, such as the steering wheel, steering shaft, electric motor, worm gear, steering gear input shaft, ball nut assembly, and output shaft. The total inertial torque must be considered. The system’s total moment of inertia \(J_{\text{total}}\) includes contributions from each component, simplified as:
$$J_{\text{total}} = J_1 + J_2 + J_3 + J_4 + J_5 + J_6 + J_7$$
where \(J_1\) to \(J_7\) represent the moments of inertia or equivalent inertial masses of the respective parts. The inertial torque \(T_{\text{inertial}}\) is calculated using the formula from electromechanical system design:
$$T_{\text{inertial}} = J_{\text{total}} \cdot \epsilon_3$$
with \(\epsilon_3\) as the angular acceleration during motor operation. The actual required motor torque \(T_{\text{motor}}\) must satisfy both the load torque \(T_{\text{load}}\) and inertial torque, adjusted for mechanical efficiency \(\eta\):
$$T_{\text{motor}} = \frac{T_{\text{load}} + T_{\text{inertial}}}{\eta}$$
Substituting values from the vehicle parameters, we find \(T_{\text{motor}} = 0.742 \, \text{N·m}\), which exceeds the 0.7 N·m threshold. This indicates that the inertial torque pushes the system over the limit, causing the EPS to exit autonomous mode. The motor control unit plays a pivotal role here, as it monitors torque and triggers mode changes based on the threshold.
To rule out motor-related issues, we performed matching calculations for overheating and overload. The actual power requirement is:
$$P_{\text{actual}} = \frac{T_{\text{motor}} \times n_{\text{max}} \times \gamma}{9.55 \times 20}$$
where \(n_{\text{max}}\) is the maximum motor speed, and \(\gamma\) is a power coefficient (1.2 to 2.5). Using \(\gamma = 2.5\), we get \(P_{\text{actual}} = 185 \, \text{W}\). The equivalent torque \(T_n\) for variable loads is:
$$T_n = \sqrt{\frac{T_1^2 t_1 + T_2^2 t_2 + \cdots + T_n^2 t_n}{t_1 + t_2 + \cdots + t_n}}$$
which yields \(T_n = 0.62 \, \text{N·m}\). The motor’s non-overheating condition requires \(T_{\text{actual}} \leq T_{\text{rated}}\) and \(P_{\text{actual}} \leq P_{\text{rated}}\). With \(T_{\text{rated}} = 14.18 \, \text{N·m}\) and \(P_{\text{rated}} = 800 \, \text{W}\), the motor does not overheat. For overload, the condition is:
$$\frac{T_n}{T_{\text{actual}}} \leq K_m$$
where \(K_m\) is the overload coefficient (1.5). Since \(T_n / T_{\text{actual}} = 0.044\), well below 1.5, the motor does not overload. Thus, the root cause is the insufficient torque threshold setting in the motor control unit, not motor capability issues.
To address this, we optimized the control logic by adjusting the motor torque threshold from 0.7 N·m to 0.8 N·m. This change ensures that during normal autonomous driving, the torque remains within the threshold, preventing automatic exits, while still allowing prompt motor intervention during hydraulic failure. The updated threshold balances performance and safety, as the motor control unit can now accurately distinguish between normal high-torque scenarios and actual failure modes.
After implementation, the improved vehicles have accumulated over 200,000 km of operation with no reported EPS automatic exit failures. The table below summarizes key parameters and calculations involved in the analysis:
| Parameter | Symbol | Value | Description |
|---|---|---|---|
| Motor Torque Threshold | \(T_{\text{threshold}}\) | 0.7 N·m (original), 0.8 N·m (optimized) | Threshold for triggering hydraulic failure mode |
| Required Steering Gear Torque | \(T_{\text{max}}\) | 8.5 N·m | Torque to overcome ground resistance |
| Motor Assistance Torque (Calculated) | \(T_{\text{motor}}\) | 0.742 N·m | Including inertial effects |
| Front Axle Load | \(G\) | Assumed based on vehicle data | Load on front axle for resistance calculation |
| Friction Coefficient | \(u\) | 0.7 | Tire-ground friction |
| Transmission Efficiency | \(\eta_1, \eta_2\) | 0.8 each | Efficiencies of linkages |
| Motor Rated Power | \(P_{\text{rated}}\) | 800 W | Maximum continuous power |
| Overload Coefficient | \(K_m\) | 1.5 | Motor overload capacity factor |
Another table illustrates the components contributing to inertial torque, highlighting the need for comprehensive dynamics analysis in EPS design:
| Component | Moment of Inertia / Equivalent | Role in System |
|---|---|---|
| Steering Wheel | \(J_1\) | Driver input interface |
| Steering Shaft | \(J_2\) | Transmits torque to steering gear |
| Electric Motor | \(J_3\) | Provides assistance via motor control unit |
| Worm Gear | \(J_4\) | Reduction mechanism part |
| Steering Gear Input Shaft | \(J_5\) | Integrates worm wheel and screw |
| Ball Nut Assembly | \(J_6\) (mass \(m_6\)) | Converts rotary to linear motion |
| Output Shaft and Pitman Arm | \(J_7\) | Delivers torque to linkages |
The optimization demonstrates that by refining the control logic in the motor control unit, we can eliminate false triggers while maintaining safety. This involves not just threshold adjustment but also enhanced sensor integration and real-time torque monitoring. For instance, the motor control unit can employ adaptive algorithms to filter out transient inertial spikes, ensuring stable operation during aggressive steering maneuvers common in autonomous driving.
Furthermore, the EPS’s reliability hinges on the seamless coordination between hydraulic and electric systems. The motor control unit must constantly evaluate conditions such as vehicle speed, steering angle, and torque demand to optimize assistance. In autonomous mode, it processes external signals from the vehicle’s central controller, translating them into precise motor commands. Any discrepancy, like exceeding the torque threshold, can cause disengagement, underscoring the need for accurate modeling and calibration.
From a broader perspective, this case study emphasizes the importance of holistic design in autonomous vehicle systems. The EPS is not just a mechanical assembly but a cyber-physical system where control logic, dynamics, and electronics intersect. Future developments could involve machine learning-based torque prediction in the motor control unit, allowing for dynamic threshold adjustment based on driving conditions. Additionally, redundancy in sensors and dual motor control units could enhance fault tolerance, preventing automatic exits due to single-point failures.
In conclusion, through detailed analysis of the EPS control logic, torque requirements, and inertial effects, we identified that the automatic exit issue stemmed from an underestimated motor torque threshold. By adjusting this threshold and validating through extensive testing, we ensured the EPS remains stable during autonomous operation and responsive during hydraulic failures. This optimization underscores the critical role of the motor control unit in achieving robust and safe autonomous driving systems. Continued refinement of control strategies, coupled with advanced modeling techniques, will further enhance the reliability and performance of electric power steering in the evolving landscape of autonomous vehicles.