Skip to content
SIH Buddyby Ganeev Singh
Dev
All problem statements
SIH26126Worth consideringacceptance 3/5

Vision Based Autonomous Navigation for Unmanned Ground Vehicle for Outdoor environment

Bharat Electronics Limited · Robotics and Drones · Software

A respectable classic robotics problem with real public datasets, but all three components must work at once to show anything — decide early whether you have a physical rover, because a simulation-only run will be discounted by a BEL judge.

What it actually is

A ground robot working outdoors cannot rely on GPS — signal drops under trees, near buildings and in the field. The ask is a navigation system that uses cameras alone to find safe ground, work out where the vehicle is, and steer around obstacles to reach a destination. The three sub-problems are named explicitly.

What to build

A vision-only navigation module with three components: a lightweight traversability segmentation model separating safe ground from hazards such as rocks, ditches and vegetation in real time; a visual odometry or SLAM pipeline estimating the vehicle's pose from camera motion alone without GPS; and a local planner converting the traversability map and pose estimate into steering and speed commands that route around newly appearing obstacles toward a goal — demonstrated navigating from a start point to a destination without collision, with the perception and pose outputs visualised alongside the executed path.

Smallest thing that wins the room

Drive the vehicle from a start point to a goal across outdoor terrain with GPS disabled, showing the traversability overlay marking safe ground live, the visual odometry track building up, and the planner swerving around an obstacle placed in the path mid-run.

How crowded this one gets

A guess, projected from the 2025 statements — the last year where both the submission counts and the winners were published.

Moderate80–180 teams expectedroughly 1 in 67–155 wins it

Quieter than 74% of the 226 · #59 of 226 by expected field

A normal-sized field. Your idea has to be good, not miraculous.

Why: company-sponsored statements drew the smallest fields of all.

This is a guess, not a fact

Nobody has published 2026’s numbers yet. This is an analysed estimate from last year’s pattern, so please do not take it as the truth — check the live counter on the SIH portal before you decide anything. The range covers the middle half of likely outcomes, so one statement in two lands outside it. Entry closes at 500 ideas per statement, so no range goes past that — a statement that reaches the cap fills and shuts rather than drawing an unlimited crowd. The model reads only three things a team can see before choosing — software or hardware, the theme, and what kind of body posted it — and those explain about a quarter of the variation in last year’s field sizes (R² 0.25 on held-out statements). Trust the band more than the number, and the ordering more than either. It cannot see how good your idea is, which is the part that actually decides it.

The scores

The number is the shorthand. The line under it is the reason.

What you will be writing

  • ORB-SLAM3 or DROID-SLAM visual odometry
  • Traversability segmentation on RUGD / RELLIS-3D
  • Jetson Orin onboard inference
  • ROS 2 navigation stack + local planner
  • Gazebo or Isaac Sim outdoor scenarios
  • Stereo or monocular depth estimation
  • Autonomous ground vehicles
  • Visual SLAM
  • Robotic perception

Prior art to read before you start

GPS-denied visual navigation · off-road traversability estimation · vision-based obstacle avoidance

Analysed by Claude Opus. Every score above is a judgment call with its reasoning attached — kindly cross-check this against the official statement on the SIH portal before your team commits to it.