Allen Emmanuel Binny

I am a fourth year undergraduate student studying Vision & Intelligent Systems in the Department of the Electronics and Electrical Communication Engineering at Indian Institute of Technology, Kharagpur.

Currently, I am pursuing my Bachelor Thesis Project under the guidance of Professor Ashish R Hota on the topic of "Distributed multi-agent control for Safe Navigation using Control Barrier Functions". I am also part of the Autonomous Ground Vehicles Research Group at IIT Kharagpur, guided by Professor Debashish Chakravarty. During the Summer 2023, I had worked as a research intern at the FOCAS Lab at the Robert Bosch Centre for Cyber Physical Systems under the guidance of Professor Pushpak Jagtap.

I had spent the previous working at the Munich Institute of Robotics and Machine Intelligence, Technical University of Munich under the guidance of Dr Abdalla Swikir working on Learning of Dynamical Systems.

My research goals are to work in the field of learning-based control specifically with safety guarantees. Feel free to check my resume and drop an email if you would think I would be a good fit for your team.

Email / GitHub / Resume / LinkedIn

Experiences

	Undergraduate Researcher \| Indian Institute of Technology Kharagpur Aug '24 - Present Worked under the supervision of Professor Ashish R. Hota on developing a distributed multi-agent controller employing CBFs, utilizing saddle-point dynamics and constraint mismatch variables, enabling real-time computation on the Gazebo simulator using ROS. Currently, I am addressing challenges involving dynamic obstacles and adversarial agents to design a robust stochastic controller aimed at enhancing the safety and efficiency of multi-agent systems.
	Undergraduate Researcher \| Technical University of Munich May '24 - Present Worked under the supervision of Dr. Abdalla Swikir on learning dynamical systems in the presence of static obstacles with safety guarantees, specifically for robot motion planning, at the Munich Institute of Robotics and Machine Intelligence. We are working on using neural networks to learn dynamical systems from trajectories and providing stability guarantees using Lyapunov functions and safety guarantees using barrier certificates. This project is specifically aimed at manipulators such as the Franka Panda. I also assisted on a project involving the learning of high-dimensional demonstrations using the composition of linear-parameter varying dynamical systems, which has been submitted to ICRA 2025.
	Research Intern \| Indian Institute of Science, Bangalore May '23 - Dec '23 Worked under the supervision of Professor Pushpak Jagtap at the FOCAS Lab at the Robert Bosch Centre for Cyber-Physical Systems. I worked on symbolic control and reachability-based methods for the autonomous racing of an F1Tenth vehicle. We used the SCOTS toolbox for symbolic control, simulated using the RViz-based F1Tenth simulator. Later, we employed pFaces, a parallelized toolbox, to achieve faster computation of symbolic control. Additionally, we worked on reachability-based methods for autonomous racing of the F1Tenth vehicle.
	Undergraduate Researcher \| IIT Kharagpur May '21 - Present Worked under the supervision of Dr. Debashish Chakravarty at AGV, IIT Kharagpur on trajectory planning and control systems. I had lead a team to work on LQR control of inverted pendulum on CopellaSim for the Eyantra Competition 2022.

Projects

These include competitions, course projects and side project.

Reading Group on Safety Filters and MPC

Volunteering | AGV.AI

report /

Conducted a reading to cover the essentials of safety filters specifically on Control Barrier Functions and Model Predictive Control. This was conducted for the students of AGV.AI

Position and altitude control of quadrotor with single motor failure (GOLD)

Competition | InterIIT Tech Meet 13.0
Oct '24 - Dec '24
report /

Secured the Gold Medal among 23 teams who participated in the competition at InterIIT Tech Meet 13.0 sponsored by Ideaforge. We worked on the detection and control of a quadrotor during complete single motor failure. We employed two novel emperical based methods to detect the motor failure in the drone directly using IMU data and motor thrusts without the use of ESC data. We were able to ensure stable hover and navigation using estimator data on Gazebo simulation implemented on PX4 firmware directly. We were also able to ensure stable hover on hardware

Design and source code from Jon Barron's website