Rohit Dhakate
Papers
2
Total Citations
11
H-Index
2
About
Rohit Dhakate is a robotics researcher whose work bridges perception and control, with key contributions in robotic manipulation and cable-driven parallel robots. His most cited paper, "Depth-aware Object Segmentation and Grasp Detection for Robotic Picking Tasks" (2021, 9 citations), introduces a novel deep neural network architecture that jointly performs class-agnostic object segmentation and grasp detection using a parallel-plate gripper. A standout innovation in this work is depth-aware Coordinate Convolution (CoordConv), which significantly improves accuracy for point proposal-based grasping in cluttered environments—a critical advance for industrial picking applications. More recently, Dhakate has ventured into the control of cable-driven parallel robots (CDPRs) with his 2025 paper "CaRoSaC: A Reinforcement Learning-Based Kinematic Control of Cable-Driven Parallel Robots by Addressing Cable Sag Through Simulation" (2 citations). Here, he presents the CaRoSaC Framework, integrating realistic simulation with model-free reinforcement learning to account for cable sag—a persistent challenge in suspended CDPRs. This work demonstrates his growing expertise in combining simulation environments with learning-based control strategies. With a research trajectory spanning from perception-driven manipulation to adaptive control of complex robotic systems, Dhakate is establishing himself as a versatile roboticist whose work has practical implications for automation and manufacturing.
Research Focus
Key Achievements
Top Papers
- 1
- 2