Rohan Baijal
@rohanblueboybaijal.bsky.social
Robotics PhD @ UW | IITK, CMU
Wandering musician with a beach ball
Wandering musician with a beach ball
This project was a fun effort with Matt Schmittle, Nathan Hatch, Rosario Scalise, @mateoguaman.bsky.social, Sidharth Talia, @khimya.bsky.social, @siddhss5.bsky.social and Byron Boots.
🧵6/6
🧵6/6
April 18, 2025 at 5:56 PM
This project was a fun effort with Matt Schmittle, Nathan Hatch, Rosario Scalise, @mateoguaman.bsky.social, Sidharth Talia, @khimya.bsky.social, @siddhss5.bsky.social and Byron Boots.
🧵6/6
🧵6/6
This work is a collaboration between the Personal Robotics Lab (@siddhss5.bsky.social) and Robot Learning Lab at the University of Washington @uwrobotics.bsky.social @uwcse.bsky.social
🧵5/6
🧵5/6
April 18, 2025 at 5:56 PM
This work is a collaboration between the Personal Robotics Lab (@siddhss5.bsky.social) and Robot Learning Lab at the University of Washington @uwrobotics.bsky.social @uwcse.bsky.social
🧵5/6
🧵5/6
🤖 Real-world tested: LRN cuts down interventions on Spot and a large tracked vehicle.
✅ Plug & play: Works with nearly any local stack that accepts goal waypoints.
🔄 Auto-labeled: Trained from raw FPV videos using CoTracker to trace camera paths.
🧵4/6
✅ Plug & play: Works with nearly any local stack that accepts goal waypoints.
🔄 Auto-labeled: Trained from raw FPV videos using CoTracker to trace camera paths.
🧵4/6
April 18, 2025 at 5:56 PM
🤖 Real-world tested: LRN cuts down interventions on Spot and a large tracked vehicle.
✅ Plug & play: Works with nearly any local stack that accepts goal waypoints.
🔄 Auto-labeled: Trained from raw FPV videos using CoTracker to trace camera paths.
🧵4/6
✅ Plug & play: Works with nearly any local stack that accepts goal waypoints.
🔄 Auto-labeled: Trained from raw FPV videos using CoTracker to trace camera paths.
🧵4/6
🔥 Key insight: Robots can reason further by learning to identify distant affordable frontiers as intermediate goals.
🧠 How it works: It uses a pre-trained SAM2 backbone + small head to find frontiers in images. Given a goal, it selects the highest-scoring one to navigate to.
🧵3/6
🧠 How it works: It uses a pre-trained SAM2 backbone + small head to find frontiers in images. Given a goal, it selects the highest-scoring one to navigate to.
🧵3/6
April 18, 2025 at 5:56 PM
🔥 Key insight: Robots can reason further by learning to identify distant affordable frontiers as intermediate goals.
🧠 How it works: It uses a pre-trained SAM2 backbone + small head to find frontiers in images. Given a goal, it selects the highest-scoring one to navigate to.
🧵3/6
🧠 How it works: It uses a pre-trained SAM2 backbone + small head to find frontiers in images. Given a goal, it selects the highest-scoring one to navigate to.
🧵3/6
❗️Problem: Robots navigating with no prior maps relying only on local sensors have a limited mapping range (due to sparse/noisy depth) causing myopic decisions.
🧵2/6
🧵2/6
April 18, 2025 at 5:56 PM
❗️Problem: Robots navigating with no prior maps relying only on local sensors have a limited mapping range (due to sparse/noisy depth) causing myopic decisions.
🧵2/6
🧵2/6
Excited to attend the talk!
January 11, 2025 at 11:01 PM
Excited to attend the talk!