Sean Reynolds

Where interesting problems
and new ideas are always welcomed

Home Contact About Me Travel Projects Blog Pictures
UAV project proposal

We wish to make an RC helicopter fly without human intervention. To do this we must incorporate several inertial sensors to monitor the environment where the helicopter will be flying. The helicopter must know several factors about its surroundings in order to control itself. It must know its orientation to the earth, it must know its ground speed, and its compass heading. With these sensor readings we can accurately describe the best means for he helicopter to react in order to maintain a stable hover. We wish to give the helicopter a high level command such as fly to the next room, and for it to navigate itself. We feel that with this level of abstraction a pilots job will be several times easier. If we have an army of drones which is able to respond with this level of abstraction, they could simultaneously be controlled by one solder; much like we control the video game starcraft.

Goals This semester we wish to write high level drivers for the GPS and the compass sensor modules. We can then use their readings to accurately adjust the calibration on the accelerometer sensor. Intuetively, we figure the accelerometer should sufficiently control the helicopter through stable flight; however the issue lies in our inability to write the perfect driver. Whilst the accelerometer can measure the current orientation to the ground and we can adjust for that; we cannot know 100% that the helicopter is actually flying steady with this reading. The problem is that an accelerometer has no way of measuring current velocity. We need a way to calibrate this throughout the flight of the helicopter. As the batteries wear down the controls need to automatically be adjusted and re-calibrated. As you could imagine this significantly reduces the enjoyment or effectiveness of flying from the pilots perspective.

The accelerometer provides great readings for orientation, thus when the helicopter tips to the right, we can adjust the blades to push it to the left; keeping it steady. In order to measure current velocity we can use a GPS module. Our GPS module is able to measure groundspeed with an accuracy of .2 mph. While this is not 100% accurate or precise, this does give us a fantastic way of providing the much needed feedback to our accelerometer sensor. Along with the groundspeed, the GPS also tells us the direction the unit is traveling. While this sounds like it might be all that we need; its important to remember that the helicopter is fully able to move in 4 axis: up-down, left-right, forward-backward and most importantly spinning about the axis defined by the blades drive shaft. This extra axis means that the orientation of the helicopter is not only able to change, but will change as the helicopter takes off or the batteries wear out. This is important because it means we must also include a fixed compass which is able to give us our orientation. With the orientation adjustment, we are able to provide the feedback to the helicopter no matter which way the nose of the chopper is facing.

Communication with the UAV
Obviously it does us little good to have a helicopter that we have no control over. The objective of our project is to provide a higher level of control, and let the helicopter figure out the rest. To communicate with the helicopter we choose bluetooth. The bluetooth module gives us maximum flexibility with our ground unit. Because we are using bluetooth, we could control the uav with any device from an iphone to a laptop computer. Bluetooth is flexible. We never want to loose contact with our UAV so we will have a network of bluetooth modules communicating to each other and to the UAV. This may provide us with localization information, however we do not know what to expect as far as usefulness. One major drawback of GPS is its ineffectiveness indoors. The localization provided by the bluetooth may come in handy when tackling these types of tasks.

Future development
This is a project with 1000’s of future development options. Everything from more friendly consumer RC helicopters, which children could pilot without worry of crashing, to an army of military drones. Our project is the groundwork for the next decade of innovation with Autonomous UAV’s. We wish to generalize the configurations of our on board sensors in such a way that they are able to map its readings to the unmanned platform its attached to. This could open up a whole new arena of machine learning. This single device could have generic actuator collections, and use its IMU readings to map out desired movement to each actuator. The possibilities are endless as we open the door to an abstraction of flight which allows for higher level control of unmanned vehicles. But just for an idea here are some of the practical ways we thing a project like this can pay for itself:
  • Arial surveillance
  • Arial photography
  • Arial recognizance
  • Target acquisition and painting
  • High speed flexible pursuit
  • Entertainment
  • Research
  • Facility monitoring
Project Links: