Kasra Khosoussi

Who am I? [CV]

I am a Research Scientist in the Laboratory for Information and Decision Systems (LIDS) and the Department of Aeronautics and Astronautics (MIT AeroAstro). I am also a member of the MIT Aerospace Controls Lab. Prior to this, I was a postdoc at MIT LIDS / AeroAstro / Aerospace Controls Lab collaborating with Prof. Jonathan P. How.

I am currently serving as an Associate Editor for IEEE Robotics and Automation Letters (RA-L). I was also an Associate Editor for ICRA 2020 and IROS 2020.

I was a finalist for the ICRA 2018 Best Multi-Robot Systems Paper Award.

Education:

I received my Ph.D. (Robotics) in 2017 from the Centre for Autonomous Systems (CAS) at the University of Technology Sydney, where I was fortunate to be advised by Prof. Shoudong Huang and Prof. Gamini Dissanayake. I was also a visiting student in 2015-2016 at the University of Southern California where I was fortunate to work with Prof. Gaurav S. Sukhatme. I received my B.Sc. degree in Computer Engineering from the Department of Electrical and Computer Engineering, K. N. Toosi University of Technology, where I was fortunate to be advised by Prof. Hamid D. Taghirad (ARAS).

Selected Publications:

  • “Distributed Certifiably Correct Pose-Graph Optimization” (IEEE T-RO 2021 - cond. accepted) [pdf]
  • “Reliable Graphs for SLAM” (IJRR 2019 - Invited Paper) [pdf]
  • “A Resource-Aware Approach to Collaborative Loop Closure Detection with Provable Performance Guarantees” (IJRR 2020 - Invited Paper) [pdf]
  • “CLEAR: A Consistent Lifting, Embedding, and Alignment Rectification Algorithm for Multi-View Data Association” (IEEE T-RO 2020) [pdf]
  • “A Sparse Separable SLAM Back-End” (IEEE T-RO 2016) [pdf]

Research:

My research focuses on the mathematical and algorithmic foundations of perception for safe and reliable autonomy in the real world. Perception is the Achilles’ heel of autonomy—and for good reason: despite tremendous progress, perception algorithms are generally brittle, require ad hoc tuning, and are not amenable to verification.

These issues have created a gap between demo (proof of concept) and deployment of autonomous robots in the real world. Purely system-engineering remedies and heuristics cannot address this issue alone. Rather, reliability must stem from and be intrinsic to the underlying algorithms.

The overarching goal of my research is to bridge this reliability gap by developing trustworthy spatial perception algorithms & systems with provable performance guarantees that can be deployed in failure-prone real-world applications and safety-critical domains.

My research builds on Optimization, Riemannian Geometry, Spectral Graph Theory, Estimation Theory, Approximation Algorithms, and Distributed Computing.

Distributed Collaborative Perception

In particular, in recent years my research has been focused on developing distributed collaborative spatial perception algorithms (e.g., distributed collaborative SLAM) with provable peformance guarantees for multi-agent systems. These include:

  • Pose-graph Optimization: Certifiably correct distributed pose-graph optimization
    (IEEE T-RO 2021 - conditionally accepted) [pdf]
  • Place Recognition: Resource-aware collaborative place recognition
    (IJRR 2020 - Invited Paper) [pdf]
  • Data Association: Cycle-consistent multi-view data association
    (IEEE T-RO 2020) [pdf]

These capabilities lie at the heart of the next generation of applications involving robotic teams, fleets of autonomous cars, and smart devices (smartphones, AR/VR) where agents with limited resources (e.g., bandwidth, battery) and partial & conflicting observations require collective spatial awareness to perform complex tasks in new environments. In particular, some of these technologies were deployed and tested in a forest search & rescue research project funded by NASA where a fleet of autonomous UAVs collaboratively build a map to navigate the forest and locate lost hikers.

:newspaper: News

:page_with_curl: Mar-21 3 (out of 3) papers accepted in ICRA 2021.
:page_with_curl: Jan-21 Our IEEE Transactions on Robotics submission “Distributed Certifiably Correct Pose-Graph Optimization” is conditionally accepted.
:man_office_worker: Aug-20 I will serve as an Associate Editor for RA-L.
:page_with_curl: Jul-20 Our invited Int. Journal of Robotics Research submission “A Resource-Aware Approach to Collaborative Loop Closure Detection with Provable Performance Guarantees” is accepted.
:page_with_curl: Jun-20 Our IEEE Transactions on Robotics submission “CLEAR: A Consistent Lifting, Embedding, and Alignment Rectification Algorithm for Multi-View Data Association” is accepted.
:man_office_worker: Jan-20 I will serve as an Associate Editor for IROS 2020.
:newspaper_roll: Dec-19 New Int. Journal of Robotics Research submission “Block-Coordinate Descent on the Riemannian Staircase for Certifiably Correct Distributed Rotation and Pose Synchronization”.
:man_office_worker: Sep-19 I will serve as an Associate Editor for ICRA 2020.
:newspaper_roll: Jul-19 New IEEE Transactions on Robotics submission “CLEAR: A Consistent Lifting, Embedding, and Alignment Rectification Algorithm for Multi-View Data Association”.
:man_teacher: Jun-19 I will co-teach VNAV (now MIT 16.485) again this Fall - feel free to reach out if you have any questions.
Older news [+]