Abstract
The emergence of Connected Automated Vehicle (CAV) has enabled a variety of Cooperative Automated Driving applications. Cooperative Adaptive Cruise Control (CACC), as the prevailing longitudinal control method for CAV allowing automated vehicle platooning, is claimed to bring enormous improvements in transportation efficiency and safety. However, such benefits of CACC cannot be easily unleashed in mixed traffic where CAVs are interacting with human-driven non-CAVs.
The existing CACC cannot work effectively in mixed traffic environment due to two limitations. Firstly, when CACC vehicles follow a human-driven and/or unconnected vehicle, they fall back to Adaptive Cruise Control (ACC), which requires much longer headway and deteriorates the traffic stability. Secondly, CACC is unable to benefit connected human-driven vehicles (CHVs) which will also largely appear in the near future. The goal of this research is to address these critical limitations of CACC, by developing quasi-CACC applications that can fully utilize the benefits of vehicular connectivity and take effects in the near future. These applications are referred as “quasi--CACC” because they aim at achieving the CACC-like behaviors while the equipped vehicles do not fully meet the operating requirements, especially in mixed traffic environment.
To address the issue of unconnected vehicle in the traffic, a CACC algorithm with unconnected vehicle in the loop (CACCu) is proposed. Unlike the traditional CACC that requires a connected preceding vehicle, CACCu aims to closely follow an unconnected preceding vehicle utilizing the information from the further (connected) preceding vehicle. CACCu is designed to maintain string stability given various behaviors of unconnected vehicles, without requiring identification process or extra information on the unconnected vehicles. A linear time-invariant CACCu on top of feedback-feedforward control structure of typical CACC is first designed. It is analytically proven that by attaching a filter of “virtual preceding vehicle” to the original CACC feedforward filter, the CACCu vehicle can stay string-stable at a gap significantly shorter than that required by ACC. The proposed CACCu along with ACC and Connected Cruise Control (CCC) were evaluated in high-fidelity simulations using real vehicle trajectory data from Next Generation Simulation (NGSIM) program and a physics-based vehicle dynamics model from PreScan. Results showed that CACCu avoided most of speed overshootings happening to ACC and CCC, indicating improved string stability. CACCu also led to overall 6~9% acceleration reduction, 35~49% spacing error reduction and 3~7% fuel saving from ACC. Compared with CCC, CACCu achieved 5~8% acceleration reduction, 26~38% spacing error reduction and 2~3% fuel saving. These numbers indicated benefits of CACCu in safety, ride comfort and energy efficiency. Then, an Adaptive Model Predictive Control (A-MPC) approach is proposed to optimize the performance of CACCu. This method make use of both a priori knowledge on the human driver’s behaviors, and the real-time observation of the actual traffic situation. The simulation results indicated that this A-MPC CACCu can facilitate a more robust implementation.
Moreover, the favorable behaviors of CACCu was validated in the field with real vehicles. CACCu reduced 10.82% acceleration RMS, 60.79% spacing error RMS and 6.24% fuel consumption from ACC’s. Compared with human driving, CACCu reduced 17.64% acceleration and 13.43% fuel consumption. The speed profiles showed that CACCu greatly attenuated the traffic disturbances while ACC and human driving tended to amplify them. It was confirmed that CACCu can greatly attenuate the traffic disturbance and improve safety, comfort, and fuel efficiency.
On the other hand, a human-in-the-loop CACC algorithm (hCACC) is developed for human-driven connected vehicle. In hCACC, the human driver remains engaged in the longitudinal control of the vehicle, while hCACC controller applies additional acceleration/deceleration on top of human actions according to the received status of preceding vehicle. By allowing coexistence of the automatic control and driver’s actions in a beneficial way, hCACC helps the human driver stabilize the vehicle more efficiently and safely. The proposed hCACC inherits the feedback-feedforward control structure and velocity-dependent spacing policy from typical CACC. String stability analysis shows that hCACC can offer broader string-stable ranges of human parameters than human driving alone or the existing human-in-the loop Connected Cruise Control (CCC), indicating a better capability to mitigate traffic disturbance with the uncertain human behaviors. The desirable properties of hCACC were validated in driving simulator experiments, which showed that hCACC could reduce 36.8% acceleration, 31.2% time-gap fluctuation, 81.2% exposure time to unsafe driving situations, and 15.8% fuel consumption from those of human driving alone.
