Assignment 5: Your Robot’s Sensors#

Your drone is equipped with three sensors:

  1. An inertial measurement unit (IMU)

  2. A time-of-flight (ToF) sensor

  3. A downward facing camera.

Thanks to these sensors, the drone is equipped with enough understanding of its environment to control its flight and fly autonomously. Each sensor is described below. By interfacing with each of these sensors, you will gain exposure to core robotics concepts including frame conversions, interpreting digital signals, and computer vision.

Time-of-flight Sensor#

A range sensor is any sensor that measures the distance to an object. There are three main types that are used on quadcopters: ultrasonic, infrared, and time-of-flight. For ultrasonic and infrared, a wave is emitted from one element of the sensor and received by the other. The time taken for the wave to be emitted, reflected, and be absorbed by the second sensor allows the range to be calculated. Infrared is more accurate, less noisy, and has a better range than the ultrasonic range sensor. The time-of-flight sensor shines infrared light at the world and measures how long it takes to bounce back. Your drone uses the time-of-flight (TOF) sensor because it accurately measures range and does not require an extra analog to digital converter board as does the infrared sensor.

Inertial Measurement Unit (IMU)#

An IMU is a device that uses accelerometers and gyroscopes to measure forces (via accelerations) and angular rates acting on a body. The IMU on the Duckiedrone is a built-in component of the flight controller. Data provided by the IMU are used by the state estimator, which you will be implementing in the next project, to better understand its motion in flight. In addition, the flight controller uses the IMU data to stabilize the drone from small perturbations.

The IMU can be used to measure global orientation of roll and pitch, but not yaw. This is because it measures acceleration due to gravity, so it can measure the downward pointing gravity vector. However, this information does not give a global yaw measurement. Many drones additionally include a magnetometer to measure global yaw according to the Earth’s magnetic field, but the Duckiedrone does not have this sensor.

Note that IMUs do NOT measure position or linear velocity. The acceleration measurements can be integrated (added up over time) to measure linear velocity, and these velocity estimates can be integrated again to measure position. However, without some absolute measurement of position or velocity, these estimates will quickly diverge. To measures these properties of the drone, we need to use the camera as described below.

Camera#

Each drone is equipped with a single Arducam 5 megapixel camera. The camera is used to measure motion in the planar directions. This camera points down towards the ground to measure x, y, and yaw velocities of the drone using optical flow vectors that are extracted from the camera images. This is a lightweight task, meaning that it does not require a lot of computational effort, because these vectors are already calculated by the Raspberry Pi’s image processor for h264 video encoding. We also use the camera to estimate the relative position of the drone by estimating the rigid transformations between two images.