Route Planning and Monitoring

With one major objective being the establishment, cultivation, and/or harvesting of row crops, it would need to know where these rows need to be for planting, or where the existing ones are for cultivation or harvesting.

To plant, it would need to know;

  1. The precise lat long of the beginning and end of each straight row, or;
  2. The area to plot in, the necessary turn-around area, and row widths for the crop to plant.

For non-straight rows, it would need to know enough intermediate points or a formula that described the shape of the row.

To cultivate or harvest, it would need to know the same information as above (if strictly GPS-based), or at least the corners of the area with rows, with a reliance on vision-based row interpretation. If it had planted the rows, that saved information would be available to be utilized.

And then there is the safe route planning for ingress to the field (when equipped) and egress from the field when done for preparation for the next task.

Route planning and monitoring safety can cover the areas of (and yes, credits to Isaac Asimov for the general themes);

  1. Do no harm to people, pets, or livestock: This would most certainly require vision and/or other sensing device for collision avoidance
  2. Do no harm to non-fixed objects: Such as row covers, temporary fences, vehicles, farm implements, etc.
  3. Avoid collisions with fixed objects: Trees, telephone poles, fences, buildings, various hardware like wellheads.
  4. Avoid traversing areas where known hazards/issues have been identified, e.g., a low lying area where it has been stuck before after x inches of rain, nearby roadways or trails (related to 2), or any area identified as off-limits.
  5. It must obey orders given it by authorized human beings except where such orders would conflict with 1, 2, or 3.
  6. Plan routes favoring preferred traversal corridors, such as on-farm ‘roads’ and paths created for it. For example, the most efficient route that would cut across rows of soybean would achieve a lower score than one that skirted the edge of the field. However, if such a route could go through such a field in the direction of the rows without harming crops, that would be a viable one.

All traversals and field work can be monitored for energy consumption by conditions and configuration to continuously improve the route planning energy estimates.

Hence, route planning needs find the most efficient route that avoids fixed objects and hazards, while route monitoring needs to ensure all the above is avoided, with rerouting in the event of a collision detection risk.

1 Like

Currently we manually record GPS paths under human remote control. The paths are high density and cover any twists and turns needed. We save those paths in the database and you can use the web interface to select a path, and preview or edit them. Then you can load the selected path on to acorn. If Acorn is within a meter of the path it will start driving along it.

We will expand this soon with a server side composer that combines multiple paths and sends that to Acorn. As we get to whole-farm operation we will build an automatic planner to compose path segments together.

At some point we intend to add a 360 degree vision system to Acorn similar to the camera system on my Rover robot:

With the full camera system we will be able to do automatic row following, and we will also be able to do vision guided driving. With vision guided mapping the whole navigation system will change.

For GPS only operation I would like to add a big “stop bar” aka a large “bumper switch” on the front and rear of the machine that will activate the e-stop. Before the full four camera system we may also want to look at front and rear safety cameras, though it is hard to get provable safety from a camera system. LIDAR may be an interesting option too as prices are coming down. Provable safety from LIDAR is easier, though our biggest safety strategy is to make Acorn unlikely to hurt you.