Design and development of a low-cost autonomous feed pusher robot for dairy barns
This paper presents the design, development, and real-world validation of a low-cost, autonomous feed pusher robot for dairy barns that utilizes an Agilex tracked chassis, LiDAR and IMU sensors, and ROS 2 Navigation 2 to navigate uneven surfaces and avoid obstacles without requiring structural modifications to the facility.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine a dairy farm as a busy, slightly messy kitchen where cows are the hungry guests waiting for their meal. Usually, the food (hay or silage) gets pushed away from the edge of the feeding trough by the cows' hooves or just by time, leaving them with an empty spot to eat. To fix this, farmers used to have to walk the line and push the food back, or buy expensive, slow robots that needed special metal tracks on the floor to know where to go.
This paper describes the creation of a new, "smart" robot designed to be the ultimate kitchen helper for these farms. Here is how the authors built it and what they found out, explained in simple terms:
The Problem: The "Rigid" Robots
Think of current farm robots like a toddler learning to walk on a tightrope. They are very careful, but they need a perfectly flat floor and special metal rails (like training wheels) to stay on course. If there's a bump, a cow, or a pile of dirt, they get confused or stop. They are also slow and run out of battery quickly, like a phone that dies after one hour of heavy use.
The Solution: The "Agile" Robot
The team built a robot that acts more like a tank than a car.
- The Chassis: They chose a tracked robot (Agilex Bunker Pro) because, just like a tank, it can crawl over rough ground, bumps, and uneven floors without getting stuck. It doesn't need those expensive metal rails on the floor.
- The "Eyes" and "Brain": Instead of relying on pre-drawn lines, the robot uses a LiDAR (a laser scanner that sees the world in 3D points, like a bat using sonar) and a camera. It runs on a smart operating system (ROS 2) that acts like a GPS navigator. It builds a map of the barn in its head and knows exactly where it is, even if a cow wanders in front of it.
- The "Arm": The robot has a shovel attached to the front. Think of this as a giant, automated hand that can push the feed forward. The robot can raise and lower this shovel to step over small obstacles, like a person lifting their foot to step over a puddle.
How It Works: The "Push" Dance
The robot follows a specific routine, which the authors call a "feeding loop":
- Mapping: First, the robot drives around once to "learn" the shape of the barn and where the feeding troughs are.
- The Approach: It drives to the start of the trough.
- The Push: It lowers the shovel and drives forward, pushing the food back toward the cows. It uses its laser eyes to stay perfectly aligned with the trough, even if the path isn't straight.
- The Exit: When it reaches the end, it lifts the shovel (so it doesn't drag) and drives back to its "home" spot to recharge or wait for the next task.
The Results: Faster and Smarter
The team tested this robot in a real research farm and a laboratory. Here is what they discovered:
- Speed: The robot can move at 1.5 meters per second. The authors compare this to current robots that crawl at about 0.28 meters per second (roughly 1 km/h). This new robot is about 5 times faster.
- Endurance: It can work for 3 hours on a single charge, whereas older models often only last about 1 hour.
- Obstacle Avoidance: In the lab, they put chairs in the robot's path. The robot saw them, stopped, calculated a new path around them, and kept going. It took longer to go around the obstacle than to drive straight, but it did it safely without crashing.
- No Construction Needed: The biggest win is that the farmer didn't have to tear up the floor or install metal tracks. The robot works on the existing, slightly bumpy barn floor.
Limitations and Future Ideas
The authors admit the robot is a bit small. If the food is very wet and heavy, the robot sometimes struggles to push a huge pile all at once. They suggest that in the future, they could make the robot bigger or push smaller amounts of food more often.
They also mention that while the robot is great now, future versions could use even smarter "eyes" (3D scanning) to see the food itself, rather than just the trough, or even track the cows to know exactly how much food they need.
In short: The authors built a low-cost, tank-like robot that uses lasers to navigate a messy barn, pushes food to hungry cows much faster than current machines, and doesn't require expensive construction work to get started.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.