BEGIN:VCALENDAR
CALSCALE:GREGORIAN
VERSION:2.0
METHOD:PUBLISH
PRODID:-//Drupal iCal API//EN
X-WR-TIMEZONE:America/New_York
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
DTSTART:20070311T020000
TZNAME:EDT
TZOFFSETTO:-0400
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
DTSTART:20071104T020000
TZNAME:EST
TZOFFSETTO:-0500
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
SEQUENCE:1
X-APPLE-TRAVEL-ADVISORY-BEHAVIOR:AUTOMATIC
235071
20260422T151505Z
DTSTART;TZID=America/New_York:20260428T103000
DTEND;TZID=America/New_York:2
 0260428T120000
URL;TYPE=URI:https://www.wpi.edu/news/calendar/events/robot
 ics-engineering-masters-thesis-presentation-sai-hitesh-viswasam
Robotics Engineering Master\'s Thesis Presentation: Sai Hitesh Viswasam
Learning Humanoid Locomotion with Morphology Aware Transformers\n\n\n\n      \n      \n\n\n\nAbstract:
  Current humanoid locomotion policies trained with deep reinforcement lear
 ning are tightly coupled to the specific robot morphology they were traine
 d on, making them brittle to variations in joint friction, link mass, or l
 imb proportions. Prior transformer-based approaches rely on one-hot encodi
 ngs to distinguish joints, which carry no structural meaning and fail to t
 ransfer to unseen morphologies, leaving a critical gap in morphology-agnos
 tic generalization. This work addresses that gap by representing the robot
  as a kinematic graph and introducing two structure-aware components: Rand
 om Walk Structural Encodings (RWSE) that capture each joint's relational p
 osition within the body, and a FiLM conditioning layer that injects global
  physical properties into the transformer at inference time without retrai
 ning. Compared to one-hot baselines, this approach produces semantically r
 icher joint embeddings (validated via t-SNE) and improved locomotion perfo
 rmance on unseen morphology variants in IsaacGym experiments with the Unit
 ree G1 robot. Future work will extend evaluation to robots with varying jo
 int and link counts, characterize the limits of zero-shot transfer, and ex
 plore integration of formal behavioral specifications for reasoning about 
 policy reliability under morphology change.\nAdvisor: Professor Mahdi Aghe
 liCommittee: Professor Jing Xiao, Professor Constantinos Chamzas\nZoom: ht
 tps://wpi.zoom.us/j/92753424781\n
END:VEVENT
END:VCALENDAR