Harshavardhan Kamarthi
Data Seminar
Sim-to-Real Transfer of Robotic Control with Dynamics Randomization (ICRA 2018)
Task: Simple Object Manipulation by Robotic Arm
Sim-to-Real Transfer of Robotic Control with Dynamics Randomization (ICRA 2018)
Task: Simple Object Manipulation by Robotic Arm
Trained on variety of simulations with parameters sampled uniformly from range given above
Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World
Task: Simple Object Manipulation by Robotic Arm using image data
Task: Train a object detector to map camera image to correct coordinates w.r.t robot's camera
Randomization parameters:
Data Dreaming for Object Detection: Learning Object-Centric State Representations for Visual Imitation
Task: Train a object detector with automatic data augmentation for behaviour cloning, pose estimation
If the object detector successfully detects the object we can generate more data by changing the position of object, background of image and train on it.
EPOpt: Learning Robust Neural Network Policies Using Model Ensembles
Task: General learning algorithm for ensuring robustness
EPOpt: Learning Robust Neural Network Policies Using Model Ensembles
Closing the Sim-to-Real Loop:Adapting Simulation Randomization with Real World Experience
Closing the Sim-to-Real Loop:Adapting Simulation Randomization with Real World Experience
Fool the domain classifier by having similar intermediate features from either real world or simulation data
Using Simulation and Domain Adaptation to Improve Efficiency of Deep Robotic Grasping
Task: Robot grasping task
Using Simulation and Domain Adaptation to Improve Efficiency of Deep Robotic Grasping
DANN
GraspGAN
Using Simulation and Domain Adaptation to Improve Efficiency of Deep Robotic Grasping
Adding other domain randomization such as image noise, different camera pose, brightness also improved success rate by 1%. However, DANN and GraspGAN improved over simple randomization by 4-6%
Emergent Complexity and Zero-shot Transfer via Unsupervised Environment Design
Task: Generic RL algorithm for robust transfer
Minimax on uniform domain randomization does worse than PAIRED
You don't have to fine-tune everything on real-world, just specific modules.
Learning to Drive from Simulation without Real World Labels
Learning to Drive from Simulation without Real World Labels
Along with training GAN, make sure the latent space is consistent for both domains
Learning to Drive from Simulation without Real World Labels
Control system for driving can be modular
Learn a blindspot detector using human/oracle feedback