Manuel López

The interest in this project arises from the desire to automate the delivery of light cargo within a city using quadcopters. In general, the layout of buildings and roads is known, but the obstacles that a quadcopter could encounter in a pre-defined route are not known. This limitation could be overcome by flying high, but this strategy will reduce the range of the trips the quadcopter can undertake because of battery capacity limitations. With this project we would like to become familiar with… Show more

The interest in this project arises from the desire to automate the delivery of light cargo within a city using quadcopters. In general, the layout of buildings and roads is known, but the obstacles that a quadcopter could encounter in a pre-defined route are not known. This limitation could be overcome by flying high, but this strategy will reduce the range of the trips the quadcopter can undertake because of battery capacity limitations. With this project we would like to become familiar with the challenges encountered in using low computational power to plan and track a path that minimizes a quadcopter’s battery consumption.

We apply a Rapidly-exploring Random Tree (RTT) algorithm to the problem of navigating a 3D space with obstacles. The simulated quadcopter becomes aware of obstacles as it navigates its environment. We use a Model Predictive Control (MPC) algorithm to follow a smoothed version of the piecewise path returned by the RTT algorithm.

The simulated quadcopter successfully navigates in three-dimensions avoiding the obstacles. It is worth noting that the MPC algorithm uses a linearized version of the full 6 degree-of-freedom model to control the quadcopter; no Proportional-Integral-Differential (PID) control or the related tuning is needed. Show less

- A team design project of an autonomous, electric aircraft with the objective to locate three targets on the ground and to minimize the search time and estimation error
- Contribution in the design of control law and programming of the software
- Creator of the search strategy
- Team won competition among 5 teams

We apply our knowledge in aeronautics, controls, and optimization to design a low cost solar UAV with the ability to fly for days on end. A unique aspect of this plane is that major design decisions have been made by a full-fledged optimization and not rules of thumb.

The team is currently flight-testing the first design iteration.

As the project's founder, I'm involved in all areas of the endeavor: controls, communications, design, optimization, construction, simulation… Show more

We apply our knowledge in aeronautics, controls, and optimization to design a low cost solar UAV with the ability to fly for days on end. A unique aspect of this plane is that major design decisions have been made by a full-fledged optimization and not rules of thumb.

The team is currently flight-testing the first design iteration.

As the project's founder, I'm involved in all areas of the endeavor: controls, communications, design, optimization, construction, simulation, distributing tasks, and funding the operation. Show less

Developed a trading algorithm that takes as input a segment of the tick-by-tick history of the bid-ask values of the Euro-U.S. Dollar (EUR/USD) pair and makes the decision of buying or selling the pair with the purpose of making a profit.
A Neural Network predicts whether or not the most recent price is a local optimum in a cycle. A Markov Decision Process algorithm then decides to buy or sell the pair taking as input the Neural Network's prediction.