BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/Chicago BEGIN:DAYLIGHT DTSTART:20250309T030000 TZOFFSETFROM:-0600 TZOFFSETTO:-0500 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:CDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20251102T010000 TZOFFSETFROM:-0500 TZOFFSETTO:-0600 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:CST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20250624T232200Z UID:E5697BA1-F7C3-4F29-8869-C0F06A7D595F DTSTART;TZID=America/Chicago:20250618T120000 DTEND;TZID=America/Chicago:20250618T130000 DESCRIPTION:Reinforcement learning has gained traction in research as a mac hine learning method for the control of autonomous vehicles. Its paradigm of learning by traversing through a series of states has shown to work wel l with the dynamic nature of driving. However\, for safety reasons\, much of the research effort is performed in simulators. This work sought to uti lize a driving simulator to train an autonomous vehicle using various rein forcement learning methods\, with the goal of the vehicle to reach a desti nation while staying within the road and avoiding collisions with other ob stacles.\n\nThree methods were explored. The first utilized a Deep Q-Netwo rk (DQN) based algorithm with the state being defined by direct values and reward shaping. The second method extended upon the first\, utilizing a P roximal Policy Optimization (PPO) based algorithm. It included additions t o the reward function based on lessons learned and experimentation with a CNN based state representation. It was shown to learn to solve a driving s cenario with less needed steps\, as compared to the first method. The fina l method utilized behavioral cloning with images as its state representati on. It was shown to be a minor improvement over its baseline.\n\nTalk is l imited to US citizens. Registration is required for non-SwRI employees.\n\ nLunch will be provided.\n\nSpeaker(s): Joseph Clemmons\n\nBldg: Building 51\, 6220 Culebra Rd\, San Antonio\, Texas\, United States\, 78238 LOCATION:Bldg: Building 51\, 6220 Culebra Rd\, San Antonio\, Texas\, United States\, 78238 ORGANIZER:garrett.hall@swri.org SEQUENCE:22 SUMMARY:IEEE AESS: Exploring Reinforcement Learning Techniques for Autonomo us Vehicle Control Using the CARLA Simulator URL;VALUE=URI:https://events.vtools.ieee.org/m/472849 X-ALT-DESC:Description: <br /><p class="MsoNormal">Reinforcement learning h as gained traction in research as a machine learning method for the contro l of autonomous vehicles. Its paradigm of learning by traversing through a series of states has shown to work well with the dynamic nature of drivin g. However\, for safety reasons\, much of the research effort is performed in simulators. This work sought to utilize a driving simulator to train a n autonomous vehicle using various reinforcement learning methods\, with t he goal of the vehicle to reach a destination while staying within the roa d and avoiding collisions with other obstacles.</p>\n<p><span style="font- size: 11.0pt\; font-family: 'Aptos'\,sans-serif\; mso-fareast-font-family: Aptos\; mso-fareast-theme-font: minor-latin\; mso-bidi-font-family: Aptos \; mso-ligatures: standardcontextual\; mso-ansi-language: EN-US\; mso-fare ast-language: EN-US\; mso-bidi-language: AR-SA\;">Three methods were explo red. The first utilized a Deep Q-Network (DQN) based algorithm with the st ate being defined by direct values and reward shaping. The second method e xtended upon the first\, utilizing a Proximal Policy Optimization (PPO) ba sed algorithm. It included additions to the reward function based on lesso ns learned and experimentation with a CNN based state representation. It w as shown to learn to solve a driving scenario with less needed steps\, as compared to the first method. The final method utilized behavioral cloning with images as its state representation. It was shown to be a minor impro vement over its baseline.<br style="mso-special-character: line-break\;">< !-- [if !supportLineBreakNewLine]--><br style="mso-special-character: line -break\;"><!--[endif]--></span></p>\n<p>&nbsp\;</p>\n<p class="MsoNormal"> <span style="mso-fareast-font-family: 'Times New Roman'\; color: black\;"> Talk is limited to US citizens. Registration is required for non-SwRI empl oyees.</span></p>\n<p class="MsoNormal"><span style="mso-fareast-font-fami ly: 'Times New Roman'\; color: black\;">Lunch will be provided.</span></p> END:VEVENT END:VCALENDAR