Anton Pollak

MSc Robotics, Systems and Control student at ETH Zurich

github.com/acpoll | scholar.google.com

Current projects

Semester Project: Improving autonomous navigation for ANYmal quadruped using a gradient-based planner

• Continuation of a project from the Spring 2025 Perception and Learning for Robotics class at ETH’s Robotic Systems Lab

---

ForzaETH: Robotics Engineer Volunteer

• Student project at ETH’s Center for Project-based Learning (website)
• Currently rebuilding the mobile platform and integrating it with the existing software stack

---

---

Previous projects

Safe Exploration via Policy Priors

• Contribution to the research project at the Learning & Adaptive Systems Group, ETH Zurich
• NeurIPS workshop paper: SOOPER

Autonomous River Cleanup (ARC): Robotics Engineer Volunteer

• Student project at the Robotic Systems Lab ETH (website)
• Working on the Mobile Autonomous Recycling Container (MARC) with ROS Noetic (robot planning, computer vision, state machine)

---

Stereofog: Image-to-image machine translation enables computational defogging in real-world images

• Research internship at the University of Utah, USA
• Scholarship by the German Academic Exchange Service (DAAD)
• Paper published in Optics Express: DOI
• Further info on Github page: GitHub
• Internal lab presentation: YouTube

---

Bachelor thesis: Optimized Operation Management With Predicted Filling Levels of the Litter Bins for a Fleet of Autonomous Urban Service Robots

• Within MARBLE (Mobile Autonomous RoBot for Litter Bin Emptying) project at TU Berlin
• Paper published in IEEE Access: DOI

---

Real-World Robotics class: Construction of a Robotic Hand and Application of Imitation Learning

• Class of Fall 2024 at the Soft Robotics Lab at ETH
• 3 main tasks:
  • Hardware design of a robotic hand based on a design provided by the lab
  • Execution of teleoperation tasks with the hand mounted on a Franka Emika robotic arm
  • Application of Imitation Learning for autonomous cube sorting
• Low-cost high-sensitivity pressure sensors using Force Sensitive Resistors (FSRs)
• Imitation learning using Action Chunking Transformer (ACT) ¹
• Code cannot be shared yet due to unpublished research

---

Vision Algorithms for Mobile Robotics class: Visual Odometry Pipeline Project

• Class of Fall 2024 at the Robotics and Perception Group (RPG) at University of Zurich
• Group bonus project: Design of a visual odometry pipeline (GitHub, Final demo video: YouTube)
• Techniques used: Shi-Tomasi algorithm, SIFT, Kanade-Lucas- Tomasi (KLT) algorithm, RANSAC ²

---

Robotics Systems class: Chess-playing Robotic Arm

• During exchange semester at the University of Melbourne, Australia
• Construction of a robotic arm to manipulate pieces on a chess board & implementation from scratch in Matlab
• Link to final demo video: YouTube

---

---

^{_{Used template: https://github.com/plpxsk/primer-template/tree/main}}

1. Action Chunking Transformer paper ↩

2. Image attribution: KITTI dataset ↩