Terminal sliding mode control of automated car-following system without reliance on longitudinal acceleration information

Abstract

The framework of minimum sensor helps reduce the number of sensors in vehicle automation. This framework is a promising approach to decrease hardware costs of driverless vehicles without significantly losing desirable performance and reliability. Even though cost is not the concern, it also provides a straightforward fault-tolerant solution in case of sensor failures. This paper presents a terminal sliding mode (TSM) controller for automated car-following systems, where only the radar is additionally equipped and no acceleration information is used. The automated controller design is based on the so-called finite-time convergence concept after compensating for the nonlinearities of powertrain dynamics via an inverse model. The controller uses a terminal function with non-integer exponents to improve the convergence rate of sliding mode, and designs a matched terminal attractor as the reaching law outside of sliding mode. The newly designed TSM controller has high robustness to modeling uncertainties and external disturbances, without the issues of singularity and chattering that are usually accompanied with conventional counterparties. For nominal conditions, it is proved that the inter-vehicle state from any initial position asymptotically converges to zero. For uncertain conditions, the bounded closed-loop stability is guaranteed under some mild assumptions on preceding vehicle acceleration and road slope. The effectiveness of this controller is validated using computer simulations and road tests on a passenger car equipped with an internal combustion engine and 5-speed automatic transmission.