■ Robotics Institute · Carnegie Mellon · Pittsburgh
SriniVijay
Srinivasan Vijayarangan · Senior Project Scientist
I've spent 15+ years building the software that lets field robots see where they are and decide where to go: in the Atacama Desert, in sorghum and soybean rows, on a factory floor, and now on a rover built for the Moon.
- 15+years in the field
- 18papers
- 1patent
- PhDRobotics, CMU 2024
Move the cursor to tilt the scan · laser sweep every 6 s
Seeing in permanent shadow.
MoonRanger testbed with stereo cameras and laser stripers
Laser-stripe distance verification
MoonRanger is a lunar micro-rover from Carnegie Mellon and Astrobotic, for NASA. As Autonomy Software Lead I led the design and development of the software for complete autonomous driving and verified it in field demonstrations. I organized and trained the student teams working on perception, state estimation, navigation, health monitoring and logging.
The hard problem is permanently shadowed regions: no sunlight, so nothing for stereo cameras to match. The approach is to project a laser line and read it with the stereo pair.
PIs · Red Whittaker & David Wettergreen
▸ Student teams · status
▸ Returns · lunar papers
- IROS 2021Stereo Perception in the Dark using an Uncalibrated Line Laser
- iSAIRAS 2020Verification of Stereo Camera Calibration using Laser Stripers for a Lunar Micro-Rover
- iSAIRAS 2020Terrain Mapping and Pose Estimation for Polar Shadowed Regions of the Moon
- iSAIRAS 2020Formulation of Micro-Rover Autonomy Software for Lunar Exploration
About a terabyte of sorghum a day.
For high-throughput phenotyping of energy sorghum on ARPA-E TERRA (2016–17) I architected the software to acquire, log and process data from 40 cameras, 5 computers and environmental sensors, about a terabyte a day. I used structure from motion to reconstruct plants from noisy field data for phenotypic measurements. I was first author on the FSR 2017 paper, and our IROS 2017 paper was a Best Application Paper finalist.
Since 2023 I have led the navigation team at the AI Institute for Resilient Agriculture: crop-agnostic navigation, drone-based mapping for precise rover treks, and end-to-end barn-to-field-to-barn navigation in soybean and corn on RTK GPS. Vision-based obstacle avoidance is in progress.
- 40cameras
- 5computers
- ~1 TBper day
- IROS 2017best application finalist
Trail & sky.
- WP1Off-road ATV2015–17
- WP2VIL-SLAM, drones2017–18
- WP3Air–ground2019
On an autonomous off-road ATV (2015–17) I built localization from visual odometry and point-cloud scan matching. I organized and led the data collection and was a go-to person for all systems issues.
For visual-inertial-LiDAR localization from drones (2017–18) I built a framework to try different visual-inertial front-ends, and a microcontroller-based time-sync setup that synchronized the sensors and the computer to sub-millisecond accuracy (Stereo VIL-SLAM, IROS 2019). In 2019, for air-ground collaboration, I set up the navigation and mapping software for an autonomous ATV and ran experiments in hard off-road terrain.
Three deserts. One rover called Zoë.
Zoë is a solar-powered planetary-analog rover. On Life in the Atacama (NASA ASTEP, 2013–2016) it combined long-duration autonomy with a 1 m drill looking for life below the surface. I was first author on the field-experiments paper (FSR 2017).
For Planetary Rover Exploration for Active Spectroscopic Mapping (2018–2022) I led the rewrite of the autonomy stack, wrote the navigator for rugged terrain, and ran field experiments at Cuprite Hills, Nevada and Yellow Cat, Utah.
My PhD grew out of this work (CMU Robotics, 2019–2024, advised by David Wettergreen): Assisting Spectral Mapping Using Cameras (IROS 2023), two iSAIRAS 2024 papers on texture-driven hyperspectral synthesis and a curated spectral reflectance dataset, and the thesis Spectral Mapping using Simple Sensors. I am also a coauthor on NASA EMIT / Tetracorder imaging-spectroscopy papers (PSJ 2024, AGU 2023).
Factory, mine & OR.
Centimetre-accurate localization on an airplane-wing assembly floor, with minimal sensors and no global tracking system. Top view of the test floor.
I joined the Robotics Institute in 2012 as a Senior Research Programmer. On a Boeing airplane-wing assembly floor (2012–14) the robot had to be localized to centimetre accuracy with minimal sensors and no global tracking system. I wrote most of the code for the initial demonstration, implemented an ICP variant from the literature and compared it with the one in PCL.
In the same years: fiducial-based localization in real and mock mines with cameras and a 3D laser; a redesigned high-level executive for the Zoë exploration robot (plan execution, fault monitoring, contingencies), with student teams I led building its Android control tablet; optimized data structures for near-real-time response in a medical snake robot for remote surgery; and a simulator for heterogeneous swarm control.
Before robots.
$ calibrate --before-robots # NetApp, 2008–2012
- REPLICATIONconfig across datacentersOK
- HEAP WALKERgdb script, memory leaksOK
- STRESS TESTship-stopper issues foundOK
- BUILD TIMEbuild system rework−50%
- LINTenabledON
- PATENTcluster config, backup, recoveryUS 8,484,163
I studied computer science and engineering at Anna University in Chennai (2002–06), then came to Carnegie Mellon for an MS in Information Technology, specializing in embedded software engineering (2006–08).
Then four years at NetApp as a Member Technical Staff: early design and prototypes for a configuration replication service across datacenters, a gdb heap-walking script for memory leaks, scalability stress tests that found ship-stopper issues, and a build time cut by 50%. One US patent (2010).
Alongside NetApp I was a visiting scholar at CMU: a waypoint tool for planning a robotic Moon mission (2008) and image warping for hemispherical displays (2009). The photo is from 2008, on a class visit to a humanoid robotics lab at CMU while I was a teaching assistant.
LOG2026 → 2008 · papers · contact
Career log, papers and contact
$ tac career.log # newest first
- NOWSenior Project Scientist · MoonRanger autonomy lead
- PHDPhD Robotics, CMU · Spectral Mapping using Simple Sensors
- AIIRAPrincipal Research Analyst · agricultural navigation team lead
- MOONMoonRanger autonomy software lead · PhD begins · air–ground collaboration
- ZOEAutonomy rewrite · rugged-terrain navigator · Cuprite Hills · Yellow Cat
- DRONESVisual-inertial-LiDAR localization · sub-ms sensor sync
- IROSBest Application Paper finalist · in-field plant segmentation
- FSRFirst author · subsurface science with long-duration autonomy (Life in the Atacama)
- SORGHUMARPA-E TERRA · 40 cameras · 5 computers · ~1 TB/day
- ATVOff-road autonomy · visual odometry + scan matching
- CMU RIMedical snake robot · swarm simulator · planetary odometry
- ZOELife in the Atacama (2013–16) · long-duration autonomy · 1 m drill · segmentation survey on Atacama data
- ZOEHigh-level executive · Android control tablet
- CMU RISenior Research Programmer · Boeing wing-floor localization · mine fiducials
- PATENTUS 8,484,163 · Cluster configuration, backup and recovery
- CMUVisiting scholar · Moon-mission waypoint tool · 2009: hemispherical display warping
- NETAPPMember Technical Staff
$ cat papers/selected.txt # 8 of 18 · first author unless marked
- iSAIRAS 2024Texture-Driven Hyperspectral Synthesis for Robotic Exploration
- iSAIRAS 2024Curated Spectral Reflectance Dataset to Aid Robotic Exploration Research
- IROS 2023Assisting Spectral Mapping Using Cameras
- IROS 2021Stereo Perception in the Dark using Uncalibrated Line Laser
- iSAIRAS 2020Verification of Stereo Camera Calibration using Laser Stripers for a Lunar Micro-Rover
- FSR 2017High-throughput Robotic Phenotyping of Energy Sorghum Crops
- FSR 2017Field Experiments in Robotic Subsurface Science with Long Duration Autonomy
- IROS 2017In-field Segmentation and Identification of Plant Structures using 3D Imaging · Best Application Paper finalist · second author
$ contact --srini
- email: enable JavaScript to show the address
- scholar
- Google Scholar →
- based
- Robotics Institute, Carnegie Mellon University, Pittsburgh
$