THIS SITE IS NO LONGER ACTIVE, GO HERE:

https://alexs28.github.io/

Dr. Alexander V. Schperberg

PhD in Robotics at the Department of Mechanical and Aerospace Engineering at the University of California Los Angeles (UCLA) - Interest in combining machine learning, controls, and vision towards locomotion and manipulation tasks
Amazon Fellow (2023)
Advised by Dr. Dennis Hong, director of the Robotics and Mechanisms Laboratory (RoMeLa)
Quick links: [CV][List of Projects][Contact]
Would you like to collaborate on my project (see 'Current Work' below) or are seeking collaboration on your own project? If so, please contact me anytime to have a chat!

Latest News:

6-13-2024: I received my PhD degree in Mechanical Engineering, specilized in Robotics. See my defense here: https://youtu.be/Gjel9Iu1CrU?si=jG0bDsyxYM8y3DcU

4-6-2024: We were accepted to the ICRA 2024 EXPO, and were selected for a travel award of 200,000 JPY (~$1,320.00). We will present the SCALER BIPED robot.

2-8-2024: Received my PhD candidacy!

1-30-2024: Our work OptiState was accepted to the IEEE International Conference on Robotics and Automation (ICRA 2024), in Yokohama Japan!

1-07-2024: Part of a patent from Mitsubishi Research Electric Laboratories (MERL) - [Patent Link]

11-19-2022: Grateful to Amazon Science for being awarded the Amazon Doctoral Fellowship! https://www.amazon.science/latest-news/amazon-and-ucla-announce-fellowship-recipients

6-8-2022: Our journal paper 'Auto-Tuning of Controller and Online Trajectory Planner for Legged Robots' has been accepted to IEEE Robotics and Automation Letters (RA-L) with the IROS option! This work was the result of my internship at Mitsubishi Research Electric Laboratories (MERL). ResearchGate Link. Will serve as co-chair at the Legged robots 3 section of IROS 2022.

8-5-2021: Our journal paper 'SABER: Data-Driven Motion Planner for Autonomously Navigating Heterogeneous Robots' has been accepted to IEEE Robotics and Automation Letters (RA-L)!

7-24-2021: Passed the 2021 PhD Preliminary Exam!

4-13-2021: Thankful and honored to be listed as a co-author on the recently published paper in the Cell Journal: 'Synthetic lethality-mediated precision oncology via the tumor transcriptome'.

2-9-2021: Will be doing a summer internship at the Mitsubishi Electric Research Laboratories (MERL) in Cambridge, Massachusetts!

8-1-2020: Our journal paper 'Machine-Learning model to predict oncologic outcomes for drugs in randomized clinical trials' was accepted to the International Journal of Cancer!

7-2-2020: Our paper 'Risk-Averse MPC via Visual-Inertial Input and Recurrent Networks for Online Collision Avoidance' was accepted to the IROS 2021 conference!

Featured work:

OptiState: State Estimation of Legged Robots using Gated Networks with Transformer-based Vision and Kalman Filtering

Summary: State estimation for legged robots is challenging due to their highly dynamic motion and limitations imposed by sensor accuracy. By integrating Kalman filtering, optimization, and learning-based modalities, we propose a hybrid solution that combines proprioception and exteroceptive information for estimating the state of the robot's trunk. Leveraging joint encoder and IMU measurements, our Kalman filter is enhanced through a single-rigid body model that incorporates ground reaction force control outputs from convex Model Predictive Control optimization. The estimation is further refined through Gated Recurrent Units, which also considers semantic insights and robot height from a Vision Transformer autoencoder applied on depth images.

Authors: Alexander Schperberg, Yusuke Tanaka, Saviz Mowlavi, Feng Xu, Bharathan Balaji, Dennis Hong

CLICK HERE TO VIEW PAPER

Auto-Tuning of Controller and Online Trajectory Planner for Legged Robots

Summary: An auto-tuning method is demonstrated on feedback controllers and online trajectory planners to achieve robust locomotion of a legged robot. Cost function weights of a Model Predictive Controller and feedback gains of a swing controller are calibrated. Further, the auto-tuning approach is used to calibrate parameters of an online trajectory planner, where the height of a swing leg and robot's walking speed are optimized.

Authors: Alexander Schperberg, Stefano Di Cairano, and Marcel Menner

CLICK HERE TO VIEW PAPER

Adaptive Force Controller for Contact-Rich Robotic Systems using an Unscented Kalman Filter

Summary: We demonstrate a self-calibrating admittance control formulation that enables wrench control (force and torques) for both manipulation and locomotion tasks. Our controller is self-calibrating using an auto-tuning method with training objectives that facilitate controller robustness/adaptability during online operation (e.g., ensuring friction cone is satisfied, spring constants are updated online to increase robustness, and reference trajectories are tracked continuously).

Authors: Alexander Schperberg, Yuki Shirai, Xuan Lin, Yusuke Tanaka, and Dennis Hong

CLICK HERE TO VIEW PAPER

Real-to-Sim: Predicting Residual Errors of Robotic Systems using Sparse Data

Summary: Achieving highly accurate kinematic or simulator models that are close to the real robot can facilitate model-based controls (e.g., model predictive control or linear-quadradic regulators), model-based trajectory planning (e.g., trajectory optimization), and decrease the amount of learning time necessary for reinforcement learning methods. Thus, the objective of this work is to learn the residual errors between a kinematic and/or simulator model and the real robot.

Authors: Alexander Schperberg, Yusuke Tanaka, Feng Xu, Marcel Menner, and Dennis Hong

CLICK HERE TO VIEW PAPER

SABER: Data-Driven Motion Planner for Autonomously Navigating Heterogeneous Robots

Summary: A data-driven motion planner for a heterogeneous multi-agent system, that moves the robots towards a goal, while considering optimality, safety, and global planning solutions (through combination of Stochastic MPC, SLAM algorithms, RNNs, and Reinforcement Learning).

Authors: Alexander Schperberg, Stephanie Tsuei, Stefano Soatto, and Dennis Hong

Click HERE for more information.

Risk-Averse MPC via Visual-Inertial Input and Recurrent Networks for Online Collision Avoidance

Summary: MPC is combined with a visual-inertial odometry SLAM (simultaneous localization and mapping) algorithm for online collision avoidance. A recurrent neural network is used to predict and propagate future uncertainty.

Authors: Alexander Schperberg*, Kenny Chen*, Stephanie Tsuei, Michael Jewett, Josh Hooks, Stefano Soatto, Ankur Mehta, and Dennis Hong

Click HERE for more information

Machine learning model to predict oncologic outcomes of drugs for randomized clinical trials

Summary: A random-forest algorithm is made to predict clinical outcomes (progression-free survival/overall survival) of cancer patients in randomized trials. Data considered to create the algorithm include genomic, transcriptomic, and proteomic information, drug-related biomarker, and clinical data.

Authors: Alexander Schperberg, Amelie Boichard, Igor Tsigelny, Stephane Richards, Razelle Kurzrock

Click HERE for more information

Current Work:

For students looking to help and collaborate on the projects below, please contact me and send me your CV.

Current project of interest:

Visual Planning, Control and Estimation for Robotic Wall Climbing

The goal is to plan the next footstep positions (online) using information from vision (i.e., map) for our latest wall climbing robot - SCALER. This project involves not only planning, but also the estimation of footsteps and control (to ensure footsteps are following its reference trajectory at all times).

Conference/Journal Peer-Review Experience:

Transactions on Robotics (T-RO 2022) - https://www.ieee-ras.org/images/publications/t-ro/2022_List_of_Reviewers.pdf
International Conference on Robotics and Automation (ICRA 2021, 2022, 2023)
International Conference on Intelligent Robots and Systems (IROS 2022, 2023)
Robotics and Automation Letters (RAL-2021, 2022)
Humanoids (2022)
Co-Chair of Legged Robot Session, IROS 2022, Kyoto, Japan

Google Sites

Report abuse