Autonomous inspection for the places no signal reaches.
No GPS, no connectivity, no pilot. The robot collects, the analysis runs on your site, and a decision-ready report lands before the morning meeting.
What happens on every run
Runs the round itself
Scheduled missions through kilometers of drifts with no GPS, no comms coverage, and nobody watching. Proven in the field, in exactly the environments where connected platforms stop working.
Compares it to the last run
The same trajectory every run means scans register cleanly. Convergence, debris, water ingress, and support damage show up as measured differences, not guesses.
Reports what needs attention
Minutes after the robot reaches comms, a report lands in the inbox: what it saw, what changed, what needs attention. No point clouds to interpret.
Run by your team. Kept on your site.
Your team writes the watch-list in plain words, places the checkpoints, approves the baseline, and sets the cadence. Changing any of it is a screen edit, not a vendor visit.
The models that read imagery and answer your checklist run on the robot and a server you host. No cloud calls, no telemetry, no training on your imagery. Air-gapped sites work.
Every round is checked against your accepted baseline. New, worse, degrading, resolved: the report says which, with the evidence attached.
Every mission adds to a dated, evidence-backed history of how the site is evolving, stored in open formats on your premises. If we part ways, you keep it all.
The report is the product
Not a dashboard to babysit, not a point cloud to decode. A document a shift can act on, delivered while the robot is still charging.
Illustrative sample data. Structure mirrors the production report; values are synthetic.
Inspection report
Sarthi Robotics · 2026-03-12
Delivered 05:41, 19 min after the robot reached comms
- Mission
- SR-2026-0312
- Type
- M2 Routine patrol with deviation monitoring
- Site
- Underground test gallery, northern Sweden
- Level
- Level 2, drift A-C loop
- Date
- 2026-03-12
- Started
- 04:10
- Reached comms
- 05:22
- Report delivered
- 05:41
- Duration
- 72 min
- Distance
- 2.1 km
- Coverage
- 96%
- Compared against
- 2026-03-10
Summary
Scheduled patrol completed without intervention. 96% of the planned route covered; one segment skipped due to standing water at chainage 1+640. Three findings flagged, one classified as new since the previous run. Gas levels nominal throughout. Recommend maintenance review of finding F-02 before the next blast cycle.
Route coverage
Environment readings
CO
4 ppm
range 0-9 ppm
NO2
0.3 ppm
range 0.1-0.6 ppm
O2
20.7 %
range 20.5-20.9 %
Temp
11.2 C
range 9.8-13.1 C
Humidity
78 %
range 71-84 %
Airflow
1.9 m/s
range 1.2-2.3 m/s
Findings (3)
Open the full sample report- F-01attentionNew since last run
New debris on floor, chainage 0+940
Comparison against the 2026-03-10 baseline run shows a debris accumulation of roughly 0.4 m height on the drift floor near the ore-pass junction. Not present in the previous scan.
Drift B, chainage 0+940
Every run compared to the last
Snapshot tools tell you what a place looks like today. Because Sarthi repeats the same route on schedule, it can tell you what moved, what grew, and what is trending toward failure.
That difference is the difference between a survey and an inspection.
Built for where the connected world ends
Most inspection robotics assumes infrastructure a mine does not have. Sarthi assumes nothing.
- Fleet platforms assume a live network to the robot.
- Sarthi's missions run entirely on the robot. The link is optional; the round happens either way.
- Survey scanners capture a site once, when someone carries them there.
- Sarthi is resident. The same route on schedule, every run compared to the last.
- Inspection robots want a pilot, a dock, or a pre-mapped facility.
- Sarthi enters unmapped workings and comes back with the map.
It lands where your team already works
Day one, the report arrives as a PDF in the inboxes you choose, in time for the morning meeting. No new software for the crew to learn, no screens to monitor.
As the routine takes hold, findings flow into the systems the mine already runs on: work orders in your maintenance system, geotech evidence for your planning tools, and a compliance trail your regulator will accept.
Proven in the field, not the lab
Years of underground trials with Nordic and European operators, and one world-first that no other stack has repeated.
World-first lava-tube exploration
Iceland
A drone explored and mapped more than 1 km of a naturally formed lava tube, fully autonomously: no GPS, no link, no pilot. The first mission of its kind anywhere.
Bring Sarthi underground
Tell us your mission profile: re-entry, routine patrol, or post-event. We will bring the field results.