Shreshta Rajakumar Deshpande’s research while affiliated with Southwest Research Institute and other places

div class="section abstract"> Onboard sensing and Vehicle-to-Everything (V2X) connectivity enhance a vehicle's situational awareness beyond direct line-of-sight scenarios. A team led by Southwest Research Institute (SwRI) demonstrated 20% energy savings by leveraging these information streams on a 2017 Prius Prime as part of the first phase of the ARPA-E-funded NEXTCAR program. Combining this technology with automation can improve vehicle safety and enhance energy efficiency further. In the second phase, SwRI demonstrated 30% energy savings over the baseline. This paper summarizes the efforts to achieve 30% savings on a 2021 Honda Clarity PHEV. The vehicle was outfitted with the SwRI Ranger automated driving suite for perception and localization. Model-based control schemes with selective interrupt and control (SIC) were used to override stock vehicle controls and actuate the accelerator, brake, and electric power steering system, enabling drive-by-wire and steer-by-wire functionalities. Key algorithms contributing to the 30% savings include Eco-driving, Eco-routing, Plugin Hybrid Electric Vehicle (PHEV) Powertrain mode selection, and cooperative maneuvers such as Eco-merge, and Platooning. These algorithms were tested through large-scale simulations using a high-fidelity forward-looking powertrain model, dynamic and stochastic traffic simulations (calibrated based on real-world corridor data), and real-world trip data. Statistical significance was established for simulation results, and a clustering and downlselection routine was used to select representative scenarios for dynamometer evaluation. This paper presents an overview of the contributing algorithms, the development of the simulation framework, the experiments designed to test the effectiveness of algorithms in simulations, an overview of the scenario downselection routine, and results from simulations and dynamometer tests. </div

Highway merges present difficulties for human drivers and automated vehicles due to incomplete situational awareness and a need for a structured (precedence, order) environment, respectively. In this paper, an unstructured merge algorithm is presented for connected and automated vehicles. There is neither precedence nor established passing order through the merge point. The algorithm relies on Control Barrier Functions for safety (collision avoidance) and for coordination that arises from exponential instability of stall-equilibria in the inter-agent space. A Monte Carlo simulation comparison to a first-in-first-out approach shows improvement in traffic flow and a significant energy efficiency benefit.

div class="section abstract"> Vehicle-to-everything (V2X) communication, primarily designed for communication between vehicles and other entities for safety applications, is now being studied for its potential to improve vehicle energy efficiency. In previous work, a 20% reduction in energy consumption was demonstrated on a 2017 Prius Prime using V2X-enabled algorithms. A subsequent phase of the work is targeting an ambitious 30% reduction in energy consumption compared to a baseline. In this paper, we present the Eco-routing algorithm, which is key to achieving these savings. The algorithm identifies the most energy-efficient route between an Origin-Destination (O-D) pair by leveraging information accessible through commercially available Application Programming Interfaces (APIs). This algorithm is evaluated both virtually and experimentally through simulations and dynamometer tests, respectively, and is shown to reduce vehicle energy consumption by 10-15% compared to the baseline over real-world routes. This paper describes the development, implementation, and validation of the algorithm. </div