Sharath Matada

I'm a senior engineer at Horizon Surgical Systems where I'm working on designing surgical robots for robotic ophthalmic surgery. I graduated from UC San Diego, where I worked with Prof. Nikolay Atanasov at the Existensial Robotics Lab. My research was focused on learning based planning.

My professional experience includes my work at Systemantics as a Senior Robotics Engineer with a focus on design and controls for 6 DOF industrial arms. I was responsible for introducing collaborative features for safe human-robot interaction. I worked under the guidance of Dr. Jagannath Raju.

Email  /  CV  /  Twitter  /  Github

Academic Research

I'm interested in learning based control and robot manipulation. I have a wide experirence in motion control and planning for robotic arms and quadrupeds.

Planning using Neural Operators

arXiv preprint  /  OpenReview

In this work, we introduce a planning neural operator (PNO) for predicting the value function of a motion planning problem. We design neural operators that map environments as cost functions to value functions. This reformulation enables us to generalize to new environments without re-trainning and inherits the super resolution properties of Fourier Neural Operators.

Pose Estimation of a dancing spot

project page  /  code    ; Reconstruction of complex robot motion from video of dynamic robot motions using keypoint detection. My work focused on integrating dynamics and simulating estimated poses on real robots.

Industrial Research and Development

I previously worked as a robotics engineer at Systemantics focusing on the design and development of India's first collaborative arm. My work here ranged from the design of six-axis serial manipulators to the design of innovative six-axis hybrid manipulators. I have worked extensively on the design and analysis, motion control, joint controller software design of these robots.

• Designed feedback controller (PID) and modeled friction, inertial, and gravity effects.
• Worked on flexible joint control using full-state feedback.

Worked on integrating collaborative features such as collision detection for safe human-robot collaboration. Used momentum based disturbance observers for flexible joints without using external force-torque sensor or internal joint torque sensors.

Worked on implementing lead through teaching using admittance controller that shapes inertia of the system.

The 6 DOF hybrid manipulator is a patented manipulator design which combines properties of both serial link and parallel link manipulators to achieve lower power consumption and large working volume

• Worked on precision design of wrist mechanism of the novel hybrid linkage consisting of six passive joints for a payload of 10 kg.
• Proposed novel alternative kinematic architecture for improved dexterity.
• Analyzed the impact of joint elasticity on the manipulator and various vibration modes resulting out of the same.
• Worked on flexible joint control for the robot joint.

Robotic manipulation tasks using an Antropomorphic End Effector

Based on Jon's website template.