Lorin Achey

Research

I work on multi-modal perception challenges for autonomous vehicles and mobile robots.

Read more about my work on risk-sensitive locomotion and navigation, generative occupancy for exploration, and multi-agent coordination under communication constraints. You can also see my talks and projects.

Research threads

Risk-sensitive locomotion and navigation In preparation

This work uses distributional reinforcement learning, which models the whole distribution of outcomes rather than just the expected return, to learn navigation and locomotion policies for quadruped agents that account for risk directly.

A quadruped robot standing on simulated rough terrain, surrounded by a grid of sampled height points
A quadruped sampling terrain around it in simulation while training with distributional reinforcement learning.

Generative occupancy for exploration

Robots normally plan only over geometry they have directly measured, so they stall when they turn a corner or enter a new room. This thread trains diffusion models to predict the geometry that has not been seen yet, fuses those predictions into a live occupancy map without ever overwriting observed space, and uses the result to keep exploring instead of pausing.

Watch Generative occupancy mapping running live on a Spot robot.
A Boston Dynamics Spot robot mapping an indoor space, with the predicted occupancy map alongside

IROS2024

SceneSense: Diffusion Models for 3D Occupancy Synthesis from Partial Observation

Alec Reed, Brendan Crowe, Doncey Albin, Lorin Achey, Bradley Hayes, Christoffer Heckman

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

Predicts occluded and out-of-view geometry in real time from a single RGB-D camera, without ever overwriting observed space.

Diffusion pipeline predicting 3D occupancy at the exploration frontier

ICRA2025

Online Diffusion-Based 3D Occupancy Prediction at the Frontier with Probabilistic Map Reconciliation

Alec Reed, Lorin Achey, Brendan Crowe, Bradley Hayes, Christoffer Heckman

IEEE International Conference on Robotics and Automation (ICRA)

73% faster runtime with minimal accuracy loss, extending prediction across the whole map rather than just around the robot.

Autonomous Robots2026

Robust Robotic Exploration and Mapping Using Generative Occupancy Map Synthesis

Lorin Achey, Alec Reed, Brendan Crowe, Bradley Hayes, Christoffer Heckman

Autonomous Robots (Springer)

Deployed on a quadruped in real-world experiments: 24% better map fidelity around the robot and 76% better at range.

Multi-agent coordination under communication constraints

When a team of robots explores somewhere the radio does not reach, the hard problem stops being perception and becomes deciding where and when to transmit. This work plans transmission locations from signal strength and payload size, so agents share what they have found without backtracking to meet.

Watch A sixty-second summary of the multi-agent exploration work.
Two robots exploring a subterranean course during a timed multi-agent run

In review2026

RF-Modulated Adaptive Communication Improves Multi-Agent Robotic Exploration

Lorin Achey, Breanne Crockett, Christoffer Heckman, Bradley Hayes

Across 480+ simulated cave environments, cut distance travelled by up to 58% and exploration time by up to 52%.

Talks

Projects

Title frame of the IsaacSim, Docker and ROS2 tutorial video

IsaacSim with Docker, ROS2 and Octomap

A walkthrough of the simulation setup I use for research: NVIDIA IsaacSim running in Docker alongside ROS2 and Octomap, with a companion repository to clone and run.

Sim-to-Sim Transfer Framework

A simulation-to-simulation transfer pipeline for validating control policies across physics engines, comparing IsaacSim with MuJoCo.

PaliGemma from Scratch

A reimplementation of the PaliGemma vision-language model architecture in PyTorch, built end to end to work through multimodal integration and large-model design.