Sensors and Sensing Lab 3: Case Studies Presentation Due on 11.01.2016, Report Due on 18.01.2016 Todor Stoyanov November 30, 2015 For this lab every group is assigned a particular case study. Follow your case study instructions and prepare a 20 minute presentation of your results. Your grade (out of 20 points) will be assigned based on your oral presentation (10 points) and the accompanying written report (10 points). The cases in this lab are more advanced and represent closer what you might face in a real world situation. Therefore, do not worry too much if you cannot find the perfect solution to each problem: it is sufficient to show that you have researched and analyzed all aspects of the problem and have some ideas of how to solve them. You can obtain up to 5 bonus points for exceptional case study presentation and full solutions.

1

Case 1: AR marker EKF SLAM

In this task you will use the ASUS Xtion Pro in a simple EKF simultaneous localization and mapping system. You will: • Use an off-the-shelf visual odometry system • Place several AR markers in a room-sized environment • Detect AR markers in the color image and use the depth image to estimate their pose relative to the sensor • Fuse observations in an Extended Kalman Filter • Evaluate system performance on simple use case scenarios

1

2

Case 2: Fotonic B70 Sensor

This case study is based on the Time of Flight camera Fotonic B70. The model B70 is no longer produced, but is similar to the C70 which is still in series at http://www.fotonic.com. Your task will be to: • Configure the sensor to work under ROS. Drivers for this sensor are available at AASS. • Study the sensor characteristics and perform a noise characterization of the sensor. Review the available scientific literature. • Propose and implement noise filtering techniques for the sensor and demonstrate your results.

3

Case 3: WiFi Arduino Differential Drive

This case study involves a larger amount of mechanical and programming work. You will use two DC motors and the Arduino Due board to build a simple differential drive kinematic and implement controllers and nodes for it. You will: • Build a chasis using aluminium profiles and mount the motors, wheels and arduino boards. • Configure the two motors on the motor shield. • Implement a velocity/steering angle controller. • Modify the Motor shiled to use free arduino pins (non-conflicting with the WiFi shield) • Implement a simple TCP/IP communication with a ROS node over the arduino WiFi shield. • Transmit left/right wheel velocity commands and receive encoder values. • Implement a ROS node that can control your differential drive with forward velocity and turning rate, while reporting odometry values. • Calibrate the odometry of the robot.

4

Case 4: Tilting Laser

This case study involves a larger amount of mechanical and programming work. You will use one DC motor, a Hokuyo URG laser scanner and the Arduino Due board to build a simple 3D laser scanner. You will:

2

• Build a housing for the laser and motor setup. • Configure the arduino board to read the laser scanner data. • Implement a motor controller and read encoder values. • Reconstruct the 3D laser geometry based on the encoder positions and 2D laser ranges. • Use ros serial to implement a node which publishes PointCloud2 messages. • Calibrate the 3D laser setup.

5

Case 5: ZED cam stereo camera

This case study involves evaluating and calibrating a consumer grade stereo camera. Note that you will need a computer with a good NVidia GPU in order to use this sensor. Your tasks will be: • Install and configure the camera. • Install and configure the ROS driver for the camera. • Use the stereo calibration toolbox to calibrate the two cameras and the baseline offset. Update values onto the camera firmware. • Design an experiment to evaluate the accuracy of the camera in typical indoor usage settings. • Vary capture parameters and resolution and compare results from the ROS stereo camera driver and the comercial GPU-based driver. • Investigate motion blur and electronic shutter effects.

6

Case 6: ROS on Raspberry Pi + Camera

This case study involves a lot of software configuration efforts. Your task will be to set-up a working ROS system on the Raspberry Pi 2 Model B ARM-based board. You will then write a node to access the camera of the Pi. You will demonstrate your system for a motion detection application. In particular you will: • Setup an Ubuntu 14.04 system on the ARM architecture provided by the Pi. • Setup and configure a ROS indigo installation on the Pi. • Setup a camera node either by porting the project https://github.com/fpasteau/ raspicam_node or by writing a new ROS node. 3

• Use simple background subtraction and filtering techniques to detect motion in the camera image. Publish on a topic when motion is detected.

7

Case 7: Sensor evaluation case studies

Select one of the following sensors, install and configure drivers. If no ROS support is available, you will need to write your own drivers. Otherwise, you will need to configure community supported drivers. Evaluate the performance of the sensor in terms of detection range and noise characteristics. Propose and evaluate filtering techniques. The following sensors are available (a separate case study can be done for each): • Kinect v2: http://www.microsoft.com/en-us/kinectforwindows/purchase/ • Leddar One: http://leddartech.com/en/leddar-sensor-products/leddarone • Leddar Tech Evaluation Kit: http://leddartech.com/en/leddar-sensor-products/ leddar-evaluation-kit

8

Case 8: XY plotter control

In this case study you will assemble an XY plotter robot using a kit from Makeblock. You will follow the guide in http://learn.makeblock.cc/xy2/ to set up the plotter’s basic functions and arduino interface. You will devise routines to evaluate the tracking performance of the plotter and present your findings. Finally, you will ook into the open case of designing a simple sample scanning program using the plotter and a petri dish holder: you will follow a square sample observation pattern and evaluate the control accuracy. This is an open case and depending on your progress further extensions are possible.

9

Case 9: Scortec ER-I

In this case you will be tasked with creating a new control box for an old manipulator at the lab. You will be using the hardware of the Scortec ER-I manipulator: http://www. intelitekdownloads.com/Manuals/Robots/Obsolete/100088-c\%20ER_I.pdf. You will be tasked with reverse-enigneering the motor and encoder connectors and plugging them into an Arduino motor shield set-up. You will devise position, velocity and current control for at least one joint of the robot and implement proper ROS interfaces. This is also an open case and depending on your progress there are several opportunities of extending the final project goals.

4

10

Case 10: Open Case

If you have ideas for a particular project using the sensors available at the lab, contact the instructors to establish the possibility to define your own sensor case study. We are open to suggestions.

5