research


2022_inhand

PatchGraph: In-hand tactile tracking with learned surface normals
Paloma Sodhi, Michael Kaess, Mustafa Mukadam, Stuart Anderson
IEEE Intl. Conf. on Robotics and Automation (ICRA), 2022
paper / video / code

Exploit local decompositions to track unseen objects during in-hand manipulations using vision-based tactile sensors. Leverage image translation GAN models for sim-to-real transfer.


2021_leo

Learning Energy-based models in Factor Graph Optimization
Paloma Sodhi, Eric Dexheimer, Mustafa Mukadam, Stuart Anderson, Michael Kaess
Conference on Robot Learning (CoRL), 2021
paper / video / code / project page

Learning observation models in factor graphs can be viewed as shaping cost functions in energy-based learning. This enables us to learn models end-to-end efficiently even with non-differentiable optimizers in the loop.


2021_gpr

Ground Encoding: Learned Factor Graph-based Models for Localizing Ground Penetrating Radar
Alex Baikovitz, Paloma Sodhi, Michael Dille, Michael Kaess
IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), 2021
paper / video / code

Off-the-shelf ground penetrating radars (GPR) to localize in unknown environments. Learn GPR factors in a sparse graph optimizer.

Best Conference Paper Finalist (0.4% of accepted papers)

2021_tactile_1

Learning Tactile Models for Factor Graph-based Estimation
Paloma Sodhi, Michael Kaess, Mustafa Mukadam, and Stuart Anderson
IEEE Intl. Conf. on Robotics and Automation (ICRA), 2021
paper / video / code / project page

Learn vision-based tactile observation models to be integrated within a sparse graph optimizer. Track objects during planar pushing.

Contributed talk at WiML workshop (4% of accepted posters)

2020_ics_contact

ICS: Incremental Constrained Smoothing for State Estimation
Paloma Sodhi, Sanjiban Choudhury, Joshua Mangelson, and Michael Kaess
IEEE Intl. Conf. on Robotics and Automation (ICRA), 2020
paper / video

Leverage primal-dual methods such as Augmented Lagrangian to solve for a constrained optimization objective online.


2019_ogm

Online and Consistent Occupancy Grid Mapping for Planning in Unknown Environments
Paloma Sodhi, Bing-Jui Ho, and Michael Kaess
IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), 2019
paper / video

Online occupancy map that maintains free space for planning while adapting efficiently to dynamically changing poses from SLAM.


2018_vog

Virtual Occupancy Grid Map for Submap-based Pose Graph SLAM and Planning in 3D Environments
Bing-Jui Ho, Paloma Sodhi, Pedro V. Teixeira, Ming Hsiao, Tushar Kusnur, and Michael Kaess
IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), 2018
paper

Planning using deformable local submaps that are efficient to update using a sparse graph-based SLAM optimizer.


2017_pheno_models

Robust Plant Phenotyping via Model-Based Optimization
Paloma Sodhi, Hanqi Sun, Barnabas Poczos, and David Wettergreen
IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), 2018
paper / code

Phenotyping can be formulated as an optimization in the space of plant models. Apply cross-entropy to obtain most likely samples.


2021_tactile_1

In-field segmentation and identification of plant structures using 3D imaging
Paloma Sodhi, Srinivasan Vijayarangan, and David Wettergreen
IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS), 2017
paper / video

Learn a mapping from multi-view in-field plant images to reconstructed 3D plant units segmented into different classes.

Best Application Paper Finalist (0.25% of accepted papers)

2015_gso

Vision-based localization and control for distributed mapping
Paloma Sodhi, Ashish Budhiraja, Achal Arvind, and Debasish Ghose
video / code

Modular multi-robot testbed implementing evolutionary optimization for distributed boundary coverage of a light source.


2015_gso

Autonomous leader-follower quadrotor control using GPS feedback
Paloma Sodhi, Ashish Budhiraja, and Debasish Ghose
video / code

Design, system ID and control for autonomous leader-follower quadrator control using GPS feedback.