You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
4
+
5
+
## Project Overview
6
+
7
+
This is Maks Sorokin's personal academic website built with Astro and Tailwind CSS. The site showcases robotics research, publications, and professional experience. It's a static site generator project deployed to GitHub Pages at https://initmaks.github.io.
8
+
9
+
## Key Commands
10
+
11
+
### Development
12
+
-`npm run dev` - Start development server with hot reloading
13
+
-`npm run build` - Build the production site
14
+
-`npm run preview` - Preview the built site locally
15
+
-`npm run astro` - Run Astro CLI commands
16
+
17
+
### Deployment
18
+
The site is automatically deployed to GitHub Pages when changes are pushed to the main branch.
19
+
20
+
## Architecture
21
+
22
+
### Tech Stack
23
+
-**Framework**: Astro 5.x - Static site generator with component islands
24
+
-**Styling**: Tailwind CSS 4.x with Vite plugin integration
I am an Applied Scientist at the <ahref="https://rai-inst.com/"class="text-gray-900 hover:text-gray-600 transition-colors">RAI Institute</a>, developing robot learning systems for whole-body manipulation. My work enables robots to have purposeful dynamic interactions with large objects and environments. I focus on fast and steerable policy learning - creating systems that can rapidly acquire new behaviors and generalize well.
42
+
I am a Research Scientist at the <ahref="https://rai-inst.com/"class="text-gray-900 hover:text-gray-600 transition-colors">RAI Institute</a>, developing robot learning systems for whole-body manipulation. My work enables robots to have purposeful dynamic interactions with large objects and environments. I focus on fast and steerable policy learning - creating systems that can rapidly acquire new behaviors and generalize well.
43
43
</p>
44
44
<pclass="mb-4">
45
45
Currently finishing my Robotics Ph.D. at Georgia Tech under
@@ -48,7 +48,7 @@ import Layout from '../layouts/Layout.astro';
48
48
<ahref="https://ckllab.stanford.edu/"class="text-gray-900 hover:text-gray-600 transition-colors">Dr. C. Karen Liu</a>, where I've built expertise across vision-based learning, navigation, manipulation, and computational design.
49
49
</p>
50
50
<p>
51
-
My goal is to develop robotic systems capable of learning and executing complex physical tasks with elegance - but at superhuman scale, speed, and precision.
51
+
My goal is to develop robotic systems capable of learning and executing complex physical tasks with elegance - at scale, and with superhuman speed and precision.
52
52
</p>
53
53
</div>
54
54
@@ -57,8 +57,7 @@ import Layout from '../layouts/Layout.astro';
Combining Sampling and Learning for Dynamic Whole-Body Manipulation
140
144
</a>
141
145
</h3>
142
146
<pclass="text-gray-600 mb-3">
143
-
Jan Brüdigam, Ali Adeeb Abbas, <strong>Maks Sorokin</strong>, Kuan Fang, Brandon Hung, Maya Guru, Stefan Sosnowski, Jiuguang Wang, Sandra Hirche, Simon Le Cleac'h
144
-
</p>
145
-
<pclass="text-sm text-gray-500 mb-6">
146
-
IEEE Robotics and Automation Letters (RA-L) 2024
147
+
<strong>Maks Sorokin</strong>*, Jan Brüdigam*, Brandon Hung*, Stephen Phillips, Dmitry Yershov, Farzad Niroui, Tong Zhao, <br> Leonor Fermoselle, Xinghao Zhu, Duy Ta, Tao Pang, Jiuguang Wang, Simon Le Cléac'h
We combined reinforcement learning with sampling-based algorithms to solve contact-rich manipulation tasks. While sampling-based planners can quickly find successful trajectories for complex manipulation tasks, the solutions often lack robustness. We leveraged a reinforcement learning algorithm to enhance the robustness of a set of planner demonstrations, distilling them into a single policy that can handle variations and uncertainties in real-world scenarios.
150
+
We give Spot the ability to manipulate heavy objects (15kg tires) by using its entire body. We combine sampling-based optimization with reinforcement learning to enable forceful, multi-contact manipulation that discovers strategies on the fly. The system handles objects exceeding the robot's nominal payload at near-human speeds.
Jacta: A Versatile Planner for Learning Dexterous and Whole-body Manipulation
179
+
</a>
180
+
</h3>
181
+
<pclass="text-gray-600 mb-2">
182
+
Jan Brüdigam, Ali Adeeb Abbas, <strong>Maks Sorokin</strong>, Kuan Fang, Brandon Hung, Maya Guru, Stefan Sosnowski, Jiuguang Wang, Sandra Hirche, Simon Le Cleac'h
183
+
</p>
184
+
<pclass="text-sm text-gray-500 mb-4">
185
+
IEEE Robotics and Automation Letters (RA-L) 2024
186
+
</p>
187
+
<pclass="text-gray-700 mb-4 leading-relaxed">
188
+
We combined reinforcement learning with sampling-based algorithms to solve contact-rich manipulation tasks. While sampling-based planners can quickly find successful trajectories for complex manipulation tasks, the solutions often lack robustness. We leveraged a reinforcement learning algorithm to enhance the robustness of a set of planner demonstrations, distilling them into a single policy that can handle variations and uncertainties in real-world scenarios.
0 commit comments