Excited to share our paper "SonicBoom" was accepted to RA-L and featured in IEEE Spectrum! 🎉
We propose a novel way to sensorize a robot arm using sound — localizing contact by listening with a microphone array. 🦾🔊
📰 Article: https://t.co/UtuhznUXBi
🧵👇
2️⃣ Audio-Visual Contact Classification for Tree Structures:
We gave farm robots the ability to sense contact by fusing sound with vision, making agriculture tasks safer and smarter.
Project Page: https://t.co/uZQ359ddSl
I'm at #CoRL and #Humanoids, and looking for full-time job positions. I'd love to connect and discuss opportunities!
Presenting:
1️⃣ SonicBoom: Contact Localization Using Array of Microphones
Featured in IEEE Spectrum, (Tues) TI4C.6, 2PM
2️⃣ Audio-Visual Contact Classification for Tree Structures
(Wed) WI2C.10, 10AM
1️⃣ SonicBoom:
We propose a novel way to sensorize a robot arm using sound — localizing contact by listening with a microphone array.
Project page: https://t.co/Qr7FPpAZdJ
Excited to share our paper "SonicBoom" was accepted to RA-L and featured in IEEE Spectrum! 🎉
We propose a novel way to sensorize a robot arm using sound — localizing contact by listening with a microphone array. 🦾🔊
📰 Article: https://t.co/UtuhznUXBi
🧵👇
💡 Key insight: The way we represent audio matters. Using a spectrogram often leads to overfitting. Instead, explicitly encoding phase information between microphone pairs significantly improved generalization to new surfaces like plastic or aluminum, when trained only on wood.
🤖📦 Want to move many items FAST with your robot? Use a tray. But at high speeds, objects may fall off 💥.
Introducing our new method: it hears sliding 🎧, learns dynamic friction ���, and plans time-optimized motions to transport objects 🚀.
https://t.co/oLlCDqUUix
🧵1/7
Very excited to share what @make_your_maark and I have been working on! We gave farm robots the ability to sense contact by fusing sound with vision, making agriculture tasks safer and smarter.
🧵1/8
Very excited to share what @make_your_maark and I have been working on! We gave farm robots the ability to sense contact by fusing sound with vision, making agriculture tasks safer and smarter.
🧵1/8
🤖We are introducing Grounded Task Axes (GTA) — a zero-shot skill transfer framework that enables robots to perform multi-step manipulation tasks on novel objects by generalizing modular controllers grounded through visual foundation models.
✅ No training, no demo, no finetuning — just zero-shot
🔧 Modular task-axis controllers
🎯 Grounded via visual foundation models
🧠 Generalizes across tools & objects
🥄🍳☕🔩 Scraping, pouring, screwing, insertion
Find more about our paper:
🌐 https://t.co/JPCDMbrBZo
📄 https://t.co/Pg3edtnf8e
🔊🦾 RoboAcoustics: Workshop on Acoustic Sensing and Representations for Robotics at ICRA 2025.
We have amazing speakers lined up! If you are at ICRA (or on Zoom), check out our workshop 5/19 Mon 9am-5pm in Room316!
Workshop Website: https://t.co/V8G5Ah2i6u
Excited to share that our paper, "Multi-Robot Motion Planning with Diffusion Models," has been selected for a Spotlight presentation at #ICLR2025!🔦🤖
We scale diffusion planning to dozens of robots *without* multi-robot data by using search.
Project: https://t.co/NZk8rEVW2t 🧵
Can 3D scene graphs act as effective online memory for solving EQA tasks in⚡️real-time?
Presenting GraphEQA🤖, a framework for grounding Vision Language Models using multimodal memory for real-time embodied question answering.