You are here: Home Staff Abhinav Valada

Dr. Abhinav Valada

Albert-Ludwigs-Universität Freiburg
Technische Fakultät
Autonome Intelligente Systeme
Georges-Köhler-Allee 080
D-79110 Freiburg i. Br., Germany
Office:   080-022

Phone:   +49 761 203-8025
Fax:   +49 761 203-8007

About me

I am a postdoctoral researcher in the Autonomous Intelligent Systems group of the Department of Computer Science at the University of Freiburg in Germany.

My research is at the intersection of robotics, machine learning and computer vision with a focus on tackling fundamental robot perception, state estimation and navigation problems using learning approaches in order to enable robots to reliably operate in more complex domains and real-world environments. The overall goal of my research is to develop scalable lifelong robot learning systems that continuously learn multiple tasks from what they perceive and experience by interacting with the real-world. My approach is to design deep learning algorithms that facilitate transfer of information through self-supervised multimodal and multitask learning by exploiting complementary features and cross-modal interdependencies. These techniques in turn enable robots to perceive more robustly and reason about the environment more effectively.

I received my PhD from the University of Freiburg under the supervision of Prof. Wolfram Burgard. Before coming to Freiburg, I was a co-founder and Director of Operations of Platypus LLC from 2013 to 2015, a robotic boats start-up company in Pittsburgh, USA. I was an enginner at the National Robotics Engineering Center in Pittsburgh from 2013 to 2014 and a systems/software engineer at the Field Robotics Center of CMU from 2011 to 2013. I received my MS degree in Robotics from Carnegie Mellon University in 2013 and my BTech degree in Electronics and Instrumentation from VIT University in 2010.

My Erdös numberis at most 4.

Publications and ongoing research can be found on the left panel.

Download my detailed Curriculum Vitae

Research Interests

  • Robot Learning
  • Self-Supervied Learning
  • Deep Multitask Learning
  • Autonomous Robot Navigation

Research Projects

  • FOUNT^2-Fliegendes Lokalisierungssystem für die Rettung und Bergung von Verschütteten.
  • RLDL-Robust Localization using Deep Landmark Features.
  • LifeNav-Reliable lifelong navigation for mobile robots.
  • ZAFH-AAL-Zentrum für Angewandte Forschung an Hochschulen für Ambient Assisted Living (Collaborative Center for Applied Research on Ambient Assisted Living)


Students Supervised

  • Dynamic Object Invariant Space Recovery, Borna Bešić, 2018, Master's Project (Ongoing)
  • Next Best View Planning for Autonomous Exploration and Mapping, Moritz Mohr, 2018, Bachelor's Thesis (Ongoing)
  • Autonomous Landing of Aerial Vehicles in Rubbles, Himanshu Maurya, 2018, DAAD Internship
  • Robust Multimodal Segmentation in Challenging Perceptual Conditions, Rohit Mohan, 2018, Bachelor's Thesis
  • Landmark-based Visual Localization using Deep Convolutional Neural Networks, Jay Patravali (together with Noha Radwan), 2017, Internship
  • Room Layout Estimation using Deep Convolutional Neural Networks, Louay Abdelgawad, 2017, Master's Project
  • Multimodal Localization using Deep Convolutional Neural Networks, Hanna Stellmach (together with Noha Radwan), 2017, Master's Project
  • Predicting Landing Sites in Aerial Images from Disaster Scenarios, Mayank Mittal, 2017, DAAD Internship
  • Laser-Camera Label Transfer for Semantic Segmentation, Rohit Suri, 2017, DAAD Internship
  • Semantic Segmentation of Moving Objects, Johan Vertens, 2016, Master's Thesis
  • Robust Deep Semantic Segmentation using Convoluted Mixture of Deep Experts, Ankit Dhall, 2016, DAAD Internship
  • Multimodal Vegetation Segmentation using Up-Convolutional Neural Networks, Julian Kunzelmann, 2016, Bachelor's Thesis
  • Navigational Autonomy for Nano-Quadrotors, Gonzalo Nuno Estevez, 2015, Bachelor's Thesis


Live Demos

  • Feb 2019: PhD defense - summa cum laude.
  • Sep 2018: Our FOUNT2 project has been seleted to represent the Ministry of Education and Research of Germany at SECURITY Essen 2018
  • Sep 2018: VLocNet++ which achieves state-of-the-art visual localization performance by leveraging the semantic and geometric knowledge of the environment is accepted for IEEE RA-L journal
  • Jun 2018: I will serve as the General Co-Chair of RSS Pioneers 2019
  • Jun 2018: "Incorporating Semantic and Geometric Priors in Deep Pose Regression" is accepted for RSS-LAIR
  • Jan 2018: VLocNet - the first CNN based localization method to outperform local feature based techniques is accepted for ICRA'18
  • Nov 2017: I was awarded the ISRR 2017 Doctorial Consortium Grant
  • Sep 2017: "Perspectives on Deep Multimodel Robot Learning" is accepted as one of the 16 papers selected for ISRR'17 Blue Sky Papers
  • Jul 2017: "Robust Proprioceptive Terrain Classification" is accepted for IJRR Special Issue on Robotics Research
  • Jun 2017: "SMSnet: Semantic Motion Segmentation" is accepted for IROS'17
  • Jan 2017: "AdapNet: Adaptive Semantic Segmentation" is accepted for ICRA'17
  • Oct 2016: Benjamin and I will give an invited talk on "Techniques for Reliable Robot Perception in Unstructured Environments" at IROS'16 workshop
  • Sep 2016: Adaptive Semantic Segmentation to be presented at NVIDIA GTC EUROPE 2016
  • Aug 2016: Paper accepted for IROS'16 workshop on State Estimation and Terrain Perception for All Terrain Mobile Robots
  • Jun 2016: Deep MultiSpectral Scene Understanding accepted for oral presentation at ISER'16
  • May 2016: Paper accepted for Deutsche Gesellschaft für Robotik-Tage 2016 (DGR Days 2016)
  • May 2016: Paper accepted for RSS'16 workshop on the Limits and Potentials of Deep Learning in Robotics
  • Mar 2016: Acoustics-based terrain classification selected for IJRR special issue on Robotics Research
  • Jan 2016: Two papers accepted for ICRA'16
  • Jul 2015: Acoustics based Terrain Classification accepted for oral presentation at ISRR'15
  • Jun 2015: Paper accepted for RSS'15 workshop on Model Learning for Human-Robot Communication
Benutzerspezifische Werkzeuge