Adaptive Cruise Control Breakthrough for Electric Vehicles
A groundbreaking advancement in adaptive cruise control (ACC) technology has emerged from Shanghai University of Engineering Science, offering a significant leap forward in the safety, efficiency, and comfort of distributed electric vehicles. Researchers Hu Shengli, Zhang Huanhuan, Jiang Zhongshun, and Chang Xiaoyu have developed a novel, multi-layered control strategy that intelligently adapts to complex driving environments, setting a new benchmark for intelligent driving systems. This comprehensive research, published in the prestigious journal Mechanical Science and Technology for Aerospace Engineering, details a holistic approach that addresses critical limitations of current ACC systems, particularly on challenging terrains like hills and curves.
The evolution of adaptive cruise control has been a cornerstone in the journey towards fully autonomous vehicles. From its origins as a simple speed-maintenance system, ACC has matured into a sophisticated technology capable of maintaining a safe distance from a lead vehicle, even bringing the car to a complete stop and resuming in traffic. The core of most modern ACC systems revolves around the concept of “time headway,” which is the time it takes for a following vehicle to reach the point where the lead vehicle currently is. Traditionally, this has been a fixed value, but the industry trend has shifted towards variable time headway (VTH) models, which can adjust the following distance based on relative speed and acceleration, offering a more natural and safer driving experience. While numerous control theories have been applied, including PID, fuzzy logic, and neural networks, Model Predictive Control (MPC) has gained prominence due to its ability to handle multiple, often conflicting, objectives—such as safety, comfort, and fuel economy—within a single optimization framework. The research team from Shanghai University of Engineering Science has built upon this foundation, identifying two critical gaps in existing systems that their new strategy aims to solve: inaccurate distance sensing on non-flat, non-straight roads and the computational inefficiency of MPC algorithms when using long prediction horizons.
The first and most fundamental challenge addressed by the researchers is the inherent inaccuracy of radar and lidar sensors when navigating hills and curves. These sensors measure the straight-line distance between two vehicles, which is not the same as the actual distance along the road. On an incline, the sensor’s line-of-sight distance is shorter than the true path distance a vehicle must travel. Similarly, on a curve, the chord distance measured by the sensor is less than the arc length of the road. This discrepancy can lead an ACC system to believe it is closer to the lead vehicle than it actually is, potentially causing unnecessary and uncomfortable braking, or in a worst-case scenario, a failure to maintain a safe distance if the system miscalculates and gets too close. To rectify this, the team designed a sophisticated spacing compensation strategy. For hill driving, they developed a model that calculates the true distance by projecting the sensor’s measurement onto the horizontal plane, effectively adding the missing horizontal component of the slope. For cornering, they used geometric principles based on the vehicle’s position and the known or estimated radius of the curve to convert the straight-line sensor data into the correct arc length. This real-time correction ensures that the ACC system always operates with an accurate understanding of the actual distance to the vehicle ahead, a crucial factor for both safety and smooth, confident operation. The effectiveness of this strategy was validated through simulations, where the compensated distance signal diverged from the raw sensor data as the lead vehicle entered a hill, then converged again once the following vehicle also ascended, proving the model’s accuracy and responsiveness.
With a reliable distance signal established, the researchers turned their attention to the core of the ACC system: the upper-level controller responsible for calculating the optimal acceleration or deceleration command. They chose MPC as the foundation for this controller, a natural fit for the multi-objective problem of balancing safety (minimizing distance and speed errors), comfort (limiting harsh acceleration and jerk), and energy efficiency (minimizing power consumption). However, a well-known limitation of MPC is the trade-off between prediction horizon and computational load. A longer horizon allows the controller to anticipate future events more effectively, leading to smoother and more optimal control. However, a longer horizon also means more calculations, which can strain the vehicle’s electronic control unit (ECU) and reduce the system’s real-time performance. To overcome this, the team introduced a “variable step-size discretization” method. This innovative approach divides the prediction horizon into two distinct parts. In the near-term, where immediate and precise control is most critical, a short time step is used, providing high resolution and accuracy. In the long-term, where the exact timing of a maneuver is less important than the general trend, a longer time step is employed, drastically reducing the number of calculations needed. This hybrid method allows the system to maintain a long prediction horizon for superior planning while keeping the computational burden manageable, ensuring both high precision and real-time responsiveness. This advancement is particularly important for electric vehicles, where energy efficiency is paramount, as a more accurately predicted and smoother driving profile directly translates to reduced energy consumption and extended range.
The upper-level MPC controller outputs a single, desired acceleration command. The final and equally critical component of the system is the lower-level controller, which translates this high-level command into specific actions for the vehicle’s actuators—the motors and brakes. For a distributed drive electric vehicle, which has an independent motor at each wheel, this presents a unique opportunity for optimization. The researchers designed a lower-level control system that includes an inverse longitudinal dynamics model, an inverse drive system model, and an inverse brake system model. This allows the controller to precisely calculate the total torque required to achieve the desired acceleration. A key feature is the brake/drive switching logic, which prioritizes regenerative braking using the electric motors. Only when the motor’s braking capacity is insufficient does the system engage the conventional hydraulic brakes, maximizing energy recovery and further enhancing efficiency.
The most innovative aspect of the lower-level control is the torque distribution strategy. Instead of simply splitting the total required torque equally between the four wheels, the researchers developed an algorithm that optimizes the distribution to maximize the overall driving efficiency of the vehicle. This strategy is based on the principle that electric motors have varying efficiency depending on their operating point—specifically, their torque and rotational speed. The algorithm continuously calculates the optimal torque split for the front and rear axles, and even between the left and right wheels on the same axle, to ensure that each motor is operating in its most efficient range. This optimization is constrained by physical limits, such as the maximum torque each motor can produce and the tire’s grip on the road, to ensure safety and stability. The team tested this strategy in two primary driving scenarios: straight-line cruising and cornering.
In the straight-line scenario, the simulation results were clear. When the vehicle was cruising at a constant speed, the optimized torque distribution strategy achieved a measurably higher total drive efficiency compared to a simple equal-split strategy. This difference, while seemingly small in percentage terms, represents a tangible gain in energy savings over the lifetime of the vehicle. The significance of this finding is that even in the most basic driving condition, a smart control strategy can extract additional efficiency from the powertrain. In the cornering scenario, the results were equally compelling. The algorithm successfully maintained the vehicle’s stability by adhering to the principles of the Ackermann steering geometry, ensuring that the inner and outer wheels received appropriate speeds. More importantly, despite the added complexity of turning, the optimized torque distribution again outperformed the equal-split method, demonstrating that the efficiency gains are not sacrificed for handling. The simulation showed that the total drive efficiency remained consistently higher throughout the maneuver, including during acceleration out of the corner.
The comprehensive nature of this research was validated through a series of rigorous simulations in the CarSim/Simulink environment, a standard tool in the automotive industry for virtual vehicle testing. Three distinct and challenging driving scenarios were used to test the entire ACC system. The first was a “stop-and-go” scenario on a straight road, simulating heavy traffic. The results showed that the system could smoothly accelerate and decelerate, maintaining a safe distance without harsh braking or jerky movements. The second scenario involved following a lead vehicle on a 10% incline. As the lead vehicle entered the hill, the spacing compensation strategy immediately corrected the distance signal, preventing the following vehicle from getting too close. The system then smoothly managed the increased power demand for climbing, demonstrating excellent tracking performance. The third scenario was a complex path with left and right turns on curves of different radii. The system seamlessly handled the transitions from straight to curved sections, with the torque distribution strategy adapting in real-time to the changing dynamics. Throughout all scenarios, the vehicle’s acceleration remained within comfortable limits, confirming the system’s focus on ride quality.
The implications of this research are far-reaching. For automotive manufacturers, this control strategy offers a ready-to-implement solution for enhancing the performance of their next-generation electric vehicles. The combination of improved safety on hills and curves, superior energy efficiency from optimized torque distribution, and a smoother, more comfortable ride directly addresses key consumer concerns. For the broader field of intelligent transportation, this work represents a significant step towards more reliable and capable autonomous driving systems. By ensuring that the vehicle has an accurate perception of its environment and can make computationally efficient, optimal decisions, this technology helps to build the foundation for higher levels of automation. The success of the variable step-size discretization method could also inspire similar approaches in other vehicle control systems, such as lane-keeping assist or predictive energy management.
In conclusion, the research team from Shanghai University of Engineering Science has delivered a comprehensive and highly effective solution to the challenges of adaptive cruise control for distributed electric vehicles. By tackling the problems of sensor inaccuracy, computational inefficiency, and suboptimal actuator control with a series of interconnected, intelligent strategies, they have created a system that is greater than the sum of its parts. Their work demonstrates a deep understanding of both the theoretical underpinnings of control engineering and the practical realities of vehicle dynamics. As the automotive industry continues its rapid transition to electrification and automation, research like this is not just an academic exercise; it is the essential engineering that will define the driving experience of the future. This new ACC strategy stands as a testament to the power of innovation in creating safer, more efficient, and more enjoyable transportation.
Hu Shengli, Zhang Huanhuan, Jiang Zhongshun, Chang Xiaoyu, Shanghai University of Engineering Science, Mechanical Science and Technology for Aerospace Engineering, DOI: 10.13433/j.cnki.1003-8728.20220211