Showing 1–2 of 2 results for author: Williams, K R

Trajectory Planning with Deep Reinforcement Learning in High-Level Action Spaces

Authors: Kyle R. Williams, Rachel Schlossman, Daniel Whitten, Joe Ingram, Srideep Musuvathy, Anirudh Patel, James Pagan, Kyle A. Williams, Sam Green, Anirban Mazumdar, Julie Parish

Abstract: This paper presents a technique for trajectory planning based on continuously parameterized high-level actions (motion primitives) of variable duration. This technique leverages deep reinforcement learning (Deep RL) to formulate a policy which is suitable for real-time implementation. There is no separation of motion primitive generation and trajectory planning: each individual short-horizon motio… ▽ More This paper presents a technique for trajectory planning based on continuously parameterized high-level actions (motion primitives) of variable duration. This technique leverages deep reinforcement learning (Deep RL) to formulate a policy which is suitable for real-time implementation. There is no separation of motion primitive generation and trajectory planning: each individual short-horizon motion is formed during the Deep RL training to achieve the full-horizon objective. Effectiveness of the technique is demonstrated numerically on a well-studied trajectory generation problem and a planning problem on a known obstacle-rich map. This paper also develops a new loss function term for policy-gradient-based Deep RL, which is analogous to an anti-windup mechanism in feedback control. We demonstrate the inclusion of this new term in the underlying optimization increases the average policy return in our numerical example. △ Less

Submitted 12 August, 2022; v1 submitted 30 September, 2021; originally announced October 2021.

Journal ref: IEEE Transactions on Aerospace and Electronic Systems, 59 (2023) 2513-2529

Equity Impacts of Dollar Store Vaccine Distribution

Authors: Judith A. Chevalier, Jason L. Schwartz, Yihua Su, Kevin R. Williams

Abstract: We use geospatial data to examine the unprecedented national program currentlyunderway in the United States to distribute and administer vaccines against COVID-19. We quantify the impact of the proposed federal partnership with the companyDollar General to serve as vaccination sites and compare vaccine access with DollarGeneral to the current Federal Retail Pharmacy Partnership Program. Although d… ▽ More We use geospatial data to examine the unprecedented national program currentlyunderway in the United States to distribute and administer vaccines against COVID-19. We quantify the impact of the proposed federal partnership with the companyDollar General to serve as vaccination sites and compare vaccine access with DollarGeneral to the current Federal Retail Pharmacy Partnership Program. Although dollarstores have been viewed with skepticism and controversy in the policy sector, we showthat, relative to the locations of the current federal program, Dollar General stores aredisproportionately likely to be located in Census tracts with high social vulnerability;using these stores as vaccination sites would greatly decrease the distance to vaccinesfor both low-income and minority households. We consider a hypothetical alternativepartnership with Dollar Tree and show that adding these stores to the vaccinationprogram would be similarly valuable, but impact different geographic areas than theDollar General partnership. Adding Dollar General to the current pharmacy partnersgreatly surpasses the goal set by the Biden administration of having 90% of the popu-lation within 5 miles of a vaccine site. We discuss the potential benefits of leveragingthese partnerships for other vaccinations, including against influenza. △ Less

Submitted 2 April, 2021; originally announced April 2021.