CS-ATR · 2024 · Robotics · Disaster response

When forward isn’t the only way through.

A capsule robot with perpendicular tracks, built to explore movement and balance on uneven terrain.

Cited by3 Google Scholar Checked 20 Sep 2026

Robot design render with a red capsule body, perpendicular black tracks, transparent end housings and a front auger — Design render for the CS-ATR and related landslide-navigation work, provided by Abhishek Sebastian. Test conditions and results are documented in the IEEE Access paper.

Inside the system / interactive concept

Sense the tilt. Change the motion.

01 / 03

The robot’s sensing and camera feed give the controller and operator information about its surroundings and orientation.

Control-loop illustration. This is not a live robot controller or a terrain simulation.

Evidence from the paper

Track arrangement: 90°
Perpendicular tracks separate longitudinal travel from lateral movement.
Mobility test: 35.06 s
Reported completion time on a prepared 10.28 m track, including lateral movement and a 5-inch pipe.
Stone-pellet terrain: 2–4 mm
The tested pellet-size transition exposed slippage and reduced traction.

Source: IEEE Access (2024), sections III and V. Results describe the paper’s prepared test environments. They do not establish deployment readiness in a real disaster site.

Built red and black capsule robot with perpendicular tracks on a prepared sand-and-gravel surface — The physical prototype on sand and gravel. Photograph provided by Abhishek Sebastian. View full photograph ↗

The terrain sets the problem

Rubble and confined spaces demand more than forward motion. Our capsule-shaped robot uses two tracks at right angles: one for travel along its length, the other for lateral movement and balance corrections.

Mechanics meet control

The study combines this track arrangement with PID tilt correction and a variational autoencoder for detecting unusual motion. I co-authored the work with Vinay Murali, R. Pragna, and A. Annis Fathima, connecting the mechanical design to its sensing and control system.

Off the screen, onto the ground

We evaluated the design in Simscape and tested a physical prototype on river sand, small stones, and simulated pits. Lower-traction surfaces exposed difficulties as well as capability. Those tests are the basis of the paper’s search-and-rescue proposal.