ROBOTICS · LULEÅ, SWEDEN
Postdoctoral researcher · Robotics & AI, Luleå University of Technology
Founder · Sarthi Robotics
I work on autonomy for aerial and ground robots in places where GPS, connectivity, and maps run out: underground mines, natural caves, and planetary analogs. The planners and mappers from my research fly real missions underground, and Sarthi Robotics is turning that stack into an autonomous inspection product.
FIELD HIGHLIGHT
The project I keep coming back to. Our drone flew into a natural lava tube, explored it entirely on its own, and came back with the map.
Over a kilometre of tunnel mapped across the campaign. The footage shows the flight in and the 3D reconstruction the robot brought back. As far as we know, this was the first fully autonomous exploration of a natural lava tube. Lava tubes also matter beyond Earth: they are the closest analog we have to the caves waiting under the surface of the Moon and Mars.
The main thread of my research: real-time exploration planning, traversability, and scene understanding for robots operating underground.
ERRT is a fully integrated, ROS-based planner that combines local exploration, safe path planning, and actuation into a real-time sampling framework. Validated in simulation and in real subterranean field trials.
A hierarchical graph-based framework for joint ground-aerial exploration using semantic traversability and a shared confidence metric, letting ground robots delegate unreachable regions to aerial counterparts in real subterranean missions.
A two-layered graph framework for large-scale exploration that reuses local sub-graphs and adapts to dynamic scene changes, updating a global navigation graph in real time to plan around blocked paths.
Belief scene graphs expand partial 3D scenes by computing the expectation of unseen objects, letting robots reason about what is likely beyond their current field of view.
A two-layer exploration and planning framework for autonomous MAVs enabling fast, continuous navigation in GPS-denied, multi-branched cave systems, with NMPC-based control for real-world subterranean deployment.
SPADE builds an actionable, multi-domain 3D scene-graph representation that scales path planning across large, semantically rich environments.
Horizon Europe and Horizon 2020 projects that put drones and ground robots into active underground mines, with industry partners including LKAB and Epiroc.
Field campaigns in real environments: operating mines, vertical shafts, and natural caves.
Lava tube, Iceland · Field campaign
Fully autonomous aerial exploration and mapping of a natural lava tube. More than 1 km explored on a single mission architecture, no GPS and no pilot.
Iron-ore mine, Northern Sweden · Field campaign
Autonomous mapping and inspection of rockfall zones after a roof collapse in an operating iron-ore mine, producing a clear survey of the affected area.
Vertical mine shaft · NexGen SIMS
Full autonomy in vertical shaft exploration with high-fidelity 3D reconstruction in real underground conditions.
Underground mines · Journal work
Autonomous post-blast gas-measurement runs in underground mines, letting crews re-enter on data rather than on a fixed clock.
Nordic mines, potash operations · Multi-year
Years of underground exploration and waypoint-navigation trials with Nordic mining companies, a mining-equipment manufacturer, and a European potash operator across EU- and industry-funded projects.
pre-seed · solo founder
Sarthi runs underground inspection rounds with no pilot and no infrastructure, and turns each run into a report: what it saw, what changed, and what needs attention. The first focus is Nordic underground hard-rock mining. The head of the Robotics & AI group at LTU advises.
Run underground operations, or invest in hard tech? I'd be glad to talk.
aaka.patel@hotmail.comThe planners, mappers, and control bridges behind the deployments — authored or co-authored, ROS-based, and tested in the field.
RUNS ON
WORKS WITH
Luleå University of Technology · Luleå, Sweden · 2021 – Nov 2025
Autonomous exploration and path planning for UAVs and legged robots in GPS-denied environments — underground mines, caves, and planetary analogs. The thesis work shipped as field-tested software and was validated in real subterranean deployments. Defended November 2025.
Luleå University of Technology · Luleå, Sweden · 2018 – 2020
Joint program across European partner universities covering space physics, orbital mechanics, and autonomous robotics. Thesis: design, modeling, and control of a coaxial quadrotor for the Martian environment.
UPES, Dehradun · Dehradun, India · 2014 – 2018
Focus on aerodynamics, propulsion, and control of aerial vehicles. Final-year project: design and simulation of a UAV for flight in Mars' thin atmosphere and reduced gravity.
Course teaching at Luleå University of Technology.
Supervising Master's students on autonomy, mapping, and field robotics.
Research collaboration, thesis supervision, underground operations, or field robotics in general — I'd be glad to hear from you.