Task 4. Preliminary mission strategy.
Discuss a preliminary mission plan / strategy (that will evolve in the next few weeks) that you are hoping to pursue during this quarter.
Describe the thinking and/or analysis that this strategy is based on.
A preliminary set of ideas (and what you will need to implement them) is all that is needed at this point.
The overall strategy is to leave the drones in standby at the various vertiports with the most nodal connections. Depending on the rate of new passengers, drones may migrate to a neighboring node with more nodal connections. We will aim to have our drones deplete their battery within 14.5 minutes, allowing time for a battery swap in the middle.
Drones will bid using a cost function whose parameters depend on the drone’s performance metrics, environmental factors, and passenger data from the central server. We will discuss each in detail below.
Drones will bid using a cost function whose parameters depend on the drone’s performance metrics, environmental factors, and passenger data from the central server. We will discuss each in detail below.
- Drone parameters are largely dependent on our final wing design, and the performance metrics will be determined empirically. The most important factor to consider when designing the wing is power consumption as a function of velocity P(u). Overall, this is something we would like to minimize such that the overall flight time during typical passenger conditions is 14.5 minutes, allowing enough time for a battery swap. We can gather this data from the flight logs of the drone. While this will not be accurate until we have the actual wings on the drone, it can give us an appropriate lower bound. The other factors to consider are the energy drain during the takeoff and landing, E-takeoff and E-landing respectively. If we attach weight approximately equal to the proposed weight of the wings to the quadcopter, we could test this in the drone lab to get an approximate value to use going forward.
- The main environmental factor to consider is the wind direction, as this can affect both our vector from the current position of the drone to the passenger, and the vector from the passenger to the final destination. This problem is further complicated due to the fact that the conditions can change both temporally and spatially, especially at different times of day and at different altitudes. To a first approximation, we will assume the wind is uniform across Lake Lagunita. We will use flight data across both drones, especially from the drone that flew most recently (or is currently flying), to estimate the wind based on the control inputs required to maintain desired headings. We will prefer flights where there is a tailwind aligned for the pickup of the passenger and the actual ride.
- The cost function will also depend on passenger data such as the frequency of ride requests, fride, how much the passengers weigh speed when accepting bids, w-speed, and the number of nodal connections at the passengers’ destination. For example, if f-ride is low, it would be advantageous for our drones to move to adjacent nodes with more connectivity in anticipation of a future ride.
- From preliminary power data on the unmodified drop the minimal power draw to speed ratio occurs at the maximum forward control authority in sport mode. This non-intuitively suggests that the optimal strategy may be to fly quickly.