You can download the full report here.


Team Peregrine has developed a unique solution to internal personal delivery within crowded, chaotic, and geometrically-challenging environments. Aerial vehicles are necessary to navigate these environments where traditional ground-based approaches fail. Challenging internal environments include sports stadiums, concert venues, and convention halls. Here, customers are disserviced by waiting in lines and would benefit from personal delivery.

Our original design, the OXcopter, features two large contra-rotating propellers above a standard quadrotor configuration: a hybrid design for ease of control and efficient flight. The quadrotor simplifies vehicle control. The large propellers provide efficient lift by maximizing rotor disk area per vehicle footprint. This creates significant efficiency advancements compared to other multirotor systems which use only a fraction of their footprint for lift. In fact, the OXcopter consumes 24% less electrical power than a quadrotor of the same size and weight.

The delivery system is designed to operate autonomously. The vending machine is responsible for dispensing the payload into the vehicle’s carriage. The vehicle then flies to its destination and lowers the carriage to complete the delivery.

The computation subsystem provides autonomous flight control, mapping, path planning, and obstacle avoidance. The LiDAR generates a map of the vehicle’s surroundings, and navigation is guided through waypoints. Heavy computations are performed off-board to reduce the weight of the vehicle.

Team Peregrine has designed and built a novel vehicle and achieved flight. The OXcopter has a 27” diameter footprint, small enough to fit through doorways, and weighs 3kg.

Executive Summary

Team Peregrine’s personal delivery solution was designed to operate in indoor environments that do not have an automated solution due to a variety of challenging constraints. Our original design, the OXcopter, is not only versatile enough to be a superior choice in disorganized settings but also offers high efficiency over other flight vehicles in lifting significant payloads. This has the potential to bring the convenience of personal delivery to customers where it has previously not been possible, a need that is well established in other capacities. Partnering with the Hershey Company, the proof of concept for our delivery system is the successful delivery of a chocolate bar to a customer.

The majority of the effort of this project was spent on the design, fabrication, and testing of the OXcopter, a hybrid design which consists of two large contra-rotating propellers above a standard quadrotor. Specifically, selecting the proper motors and propellers for both the OX (contra-rotating propellers) and the quad (standard quadrotor) systems and designing the contra-rotating assembly was crucial. The motors for the OX, as there were no off-the-shelf contra-rotating ones available at our scale, had to provide enough torque to generate the lift required for our vehicle and have a hole through the middle for our design to work. The tube that supported the upper motor and rotors had to be custom made in order to allow passage wires through the middle and carry current itself as the ground in the motor circuit. We selected two different propellers, twisted and rectangular, for the vehicle. Both fit in our maximum footprint of 66cm and could be inclined for experiments. The twisted blades performed better, but they were much more expensive than the rectangular blades.

One of the major unknowns of this project was how the lower rotor would behave in the wake of the upper rotor. The expert assumption was it would need to be inclined proportional to the flow rate of the incoming air. To test this, we designed a method to incline both the rectangular and twisted blades via 3D printed wedges of discrete angles. From these experiments, we determined that there was in fact no benefit to inclining the lower rotor (in fact, it decreased the system’s performance). Our experiments showed that the vehicle was able to generate the thrust required for flight. We achieved max thrust using a pair of twisted blades with no incline of the lower rotor. Our experiments also showed that the torques of the two large rotors can be balanced at a condition where they generate enough thrust to fly.

The quadrotor subsystem was designed to fit within the footprint of the OX. Because the OX assembly was designed to provide most of the lift, the motors and propellers for the quad were chosen so that the quad could lift itself (frame, motors, and propellers) as well as steer the vehicle. This subsystem also consisted of four custom landing feet that supported the vehicle during testing.

One goal of the Peregrine Parcel delivery system was that the system would not require human interaction. In working toward this goal, the carriage and vending subsystems were designed. The carriage was mainly constrained by the type of payload, weight, and safety. A winch system lowered a lightweight basket into the vending machine, where candy was dispensed, before raising it for takeoff. The vending machine consisted of eight drawers that were able to slide out for loading and would vend the requested candy bar into the carriage for delivery by the OXcopter.

In order to achieve a fully autonomous system, a series of high-level robotic tasks had to be executed. To complete an order, the vehicle must localize itself within a map of the environment and then plan a path from its current location to the requested destination all while avoiding obstacles. The algorithms required for such a task are complex and time intensive. To increase the rate at which we could process incoming data, we opted to move the majority of the computation to a grounded base station computer. For this we confirmed that the incoming data rate could be supported by existing wireless access points on the market. By the end of the project, a localization algorithm that used information gathered by the on-board sensors to calculate the vehicle’s position within its environment had been successfully implemented. This calculation was done on the base station computer, demonstrating the validity of the network’s speed.

In order to allow us to safely test the vehicle in flight, we designed and built a kill-switch board that could switch power to the motors. The board is designed to dissipate the power generated by up to 120 A of current, 40A more than required for the motors running on a 5S battery, and is activated and deactivated through a switch on the transmitter. To power the compute suite, namely the ODroid, Asus Xtion Pro, and the LiDAR, we designed and built a 5V regulator capable of providing up to 6A of current. The board also provides protection against current spikes from the motors.

A physical safety system adds extra protection for people who may be interacting with the OXcopter. The safety system was designed to be lightweight and strong. It consists of carbon fiber composite hoops serving as propeller guards that are supported by carbon fiber rods. This, along with the addition of a wire mesh across the guards, prevents unwanted contact with the rotors. The lower (quad) safety system was assembled on the vehicle, but the upper was not due to time and cost constraints.

Team Peregrine has designed and built a novel vehicle and achieved flight. The OXcopter has a 27” diameter footprint, small enough to fit through doorways, and weighs 3kg. And while a fully autonomous delivery was not achieved, the subsystems that were successfully prototyped represent the necessary early iterations of making this a reality.