Why Georgia Tech Built a Tarzan Robot That Swings Around on Wires
Generally, the term “aerial robot” is synonymous with “drone,” but there are lots of other ways for robots to avoid spending time on the ground. One of the most creative that we’ve seen recently comes from Georgia Tech Professor Jonathan Rogers, who has been working on a sloth-inspired aerial robot named Tarzan. The machine is designed to swing around on overhead wires strung above fields to monitor growing crops. And one day, it may swing around electrical wires in cities, too.
Tarzan is built with carbon fiber arms, reinforced with aluminum. Each arm has a dc motor at its base, and the shafts of those motors are coupled to each other. A bearing on the coupling shaft keeps the payload pod attached securely, while also allowing it to hang parallel to the ground irrespective of the orientation of the arms. The robot’s hands are 3D-printed grippers with embedded IR sensors that can detect when a wire passes them and trigger the grippers to close tightly around it. Tarzan is so new that it doesn’t have much of a payload yet, but eventually, the plan is to include a high resolution video camera as well as an IR camera for assessing plant health and hydration.
“We’ve been interested in the idea of long-term, persistent environmental monitoring in which plants can be monitored and treated on an individual basis,” Rogers told us. “The idea is that one or more low-cost robots can be deployed for an entire growing season in a field, ‘living with the crops’ so to speak, providing real-time monitoring and treatment in a very targeted way.” Such approach, he continues, would allow farmers to increase output by quickly identifying poorly performing crops and applying fertilizer, pesticides, or water more efficiently.
The system that Rogers is working towards (along with a team that includes Dr. Ai-Ping Hu and graduate students Evan Davies and Siavash Farzan) involves a system of wires installed on posts above each row of crops in a farmer’s field. Several robots would be deployed at the beginning of the growing season, and they would collaboratively and autonomously swing around to check out the crops with their sensors, potentially dispensing pesticides or fertilizer when and where they decide it’s necessary. Data is constantly sent back to the farmer for analysis and high-level decision making, although ideally, the system wouldn’t require the farmer to do much of anything.
While Tarzan will certainly have some competition in the agricultural robotics space from more traditional ground robots, the concept is compelling, especially if Georgia Tech can show that the overall system (including all the overhead infrastructure required) is both cost effective and useful long-term. I’m picturing a [insert appropriate collective noun for sloths here] of these robots swinging around and occasionally dispensing a little bit of fertilizer from above, which would be pretty amusing to watch. As for the other future applications, I think it’s just a matter of time before they realize that package-delivery robots swinging around power lines is definitely the way to go.
For more details, we spoke with Professor Rogers via email.
IEEE Spectrum: Why is having an aerial robot more useful than a ground robot for agricultural applications?
Jonathan Rogers: To avoid getting stuck in mud or tangled, ground robots must have large wheels, but large wheels require the robot to be big and increase energy consumption, and also risk trampling the plants themselves. We’d like to be able to guarantee that the robot won’t trample plants, and also will not get stuck or tangled. This is very hard to do with ground robots.
What are the advantages that Tarzan has over a more traditional, less dynamic cable-based mobility system? For example, either a cable-driven system that uses winches to position a sensor system in 3D space, or a robot that can traverse along suspended cables with wheels like a chairlift?
We looked at cable-driven systems, like the ones they use on ESPN for sports games. These systems only cover a football field, which is much smaller than a typical farm field. When scaling these systems up, the cables get very heavy and the motors needed become huge. It would be a big capital investment and very complicated to install. Compare that to some small rolls of wire and poles (both obtainable at a hardware store) needed for setting up our system.
We also considered a robot with wheels, or a chairlift-type setup. The problem is that such a system can only move in 1D—that is, it can only move along a single wire. Our robot can move along one wire and between wires. So if parallel wires are strung above each crop row, our robot can cover the full 2D space needed. Potentially, we can modify our robot so that it uses wheels to traverse a single wire, but can then swing to go between wires. We have thought about doing this.Photo: Georgia Tech Georgia Tech’s Tarzan robot team [from left]: Siavash Farzan, Ai-Ping Hu, Professor Jonathan Rogers, and Evan Davies
Will the robot be able to use its limbs for manipulation?
This is one possibility. Since the robot can swing down, it can potentially use its free hand to grab things (for example, pick vegetables and put them in a basket it carries along). There are lots of possibilities here, many of which we haven’t thought of yet.
What other applications can you imagine for this kind of system?
Power line inspection is another good application. We have also thought about using these systems in urban environments to traverse along telephone or power wires, providing reconfigurable surveillance sensors, air quality monitoring, or even reconfigurable traffic lights.
What are you working on next?
We are in the process of designing a second-generation robot that can swing both along and between cables. The next prototype of the robot will have “wrists” that allow us to rotate the hands 90 degrees. This is so that the robot can release one hand, fall down to a rest position, rotate itself, and swing back up to hook onto a different cable. We expect that to be ready within about four months. At the same time, this summer we are going to deploy the robot at a soybean research facility at the University of Georgia to swing and take measurements in an outdoor environment. This will be an important step in getting the robot out of the lab into a real agricultural setting.