Rama Krishna Prasad Kandukuri

bio-photo.jpeg
Computer Vision, Robotics, Deep Learning

Currently working as a Robot Software Engineer for Simulation at Magazino. I work on development and maintenance of custom simulation platforms for robots in Unity and Gazebo. Previously, Robot Software Engineer at Polybot, headed by Dr. Wieland Brendel. I have worked on distributed multi object tracking pipeline and diffusion-based policies deployed on the Spot robot.

Before that, I was a Researcher at Embodied Vision group supervised by Prof. Dr. Jörg Stückler at the Max Planck Institute for Intelligent Systems Tübingen, working at the intersection of Machine Learning and Vision-based robotics. Specifically vision-based object manipulation employing differentiable physics. I also work on using differentiable rendering for unsupervised learning of scene’s physics, 6D rigid body pose estimation from images, methods for planning and control through contacts and frictional constraints with learned physics, and vision-based tactile sensing.

I did my M.Sc. in Mechatronics Engineering at University of Siegen where I worked on robotics, optimization and computer vision. I did my master’s thesis under the supervision of Prof. Dr. Michael Möller at University of Siegen and Prof. Dr. Jörg Stückler at the Max Planck Institute for Intelligent systems, Tübingen. I did my B.Tech. in Mechanical Engineering at National Institute of Technology (NIT), Rourkela.

Read my resume, reach me at email.

news

Sep 01, 2024 Joined as Robot Software Engineer for simulation at Magazino
Mar 20, 2024 Paper published in 3DV 2024, Davos.
Jan 01, 2024 Joined as Robot Software Engineer at Polybot
Jul 01, 2021 Paper published in IJCV 2021.
Oct 01, 2020 Paper published in GCPR 2020, Tuebingen Honorable Mention.

selected publications

  1. 3DV
    title_ekfphys.png
    Physics-Based Rigid Body Object Tracking and Friction Filtering From RGB-D Videos
    Rama Krishna Kandukuri, Michael Strecke, and Joerg Stueckler
    In International Conference on 3D Vision (3DV) , 2024
  2. Physical Representation Learning and Parameter Identification from Video Using Differentiable Physics
    Rama KandukuriJan AchterholdMichael Moeller , and 1 more author
    International Journal of Computer Vision (IJCV), 2021
  3. Learning to Identify Physical Parameters from Video Using Differentiable Physics
    Rama KandukuriJan AchterholdMichael Moeller , and 1 more author
    In German Conference on Pattern Recognition (GCPR) , 2020
    GCPR 2020 Honorable Mention