190 likes | 307 Vues
In our midterm presentation for the Stratus project, we have made significant strides in hardware assembly, having completed mechanical assembly and received essential components like the APM case and power module. However, we've faced multiple hardware challenges, especially with the Wi-Fi module and the compatibility of APM with the Xbee Explorer. Our focus this semester shifts to software integration, aiming for a direct connection to the phone for displaying copter status and sending commands. We've outlined our approach towards integrating ROS with MAVLink and the development of our iOS and Android apps.
E N D
Ryan Rasmussen Maggie Krause Jiajun Yang Stratus projectMidterm Presentation
Hardware Progress • Mechanical assembly complete • Received APM case and power module last week • Connected wi-fi module in January but have encountered several problems so far
Hardware Problems • APM and Xbee Explorer not compatible without modification • Dr. Detweiler removed a diode and soldered additional resistors • First wi-fi module is no longer functional, now using spare • We are not sure how it broke, possibly hooked up incorrectly? • Initially used TCP connection, but discovered that when the connection is lost, it was very difficult to recover • Now using UDP, which requires connecting to module with static IP • Several connection problems when working in NIMBUS Lab • Sponsors said they have always encountered wi-fi problems in the building • We don’t have the same problems when using outside the lab
Progress • First semester was focused on assembly and choosing communication hardware • This semester is focused on software with two major goals • Direct connection to phone, which will display copter status and send basic commands • Adding node to existing NIMBUS framework to allow the lab’s ROS system to communicate with the copter’s MAVlink system • Separation of tasks • NIMBUS Integration – Maggie • iOS App – Jiajun • Android App - Ryan
NIMBUS Integration: ROS Overview • Robot Operating System (ROS) • Open-source "meta-operating system" for communicating with robots • Built to be modular and adaptable • Provides a host of command-line tools for testing and development • rosmake, rosrun, roslaunch, rxgraph, etc.
ROS Concepts • Uses a Computation Graph: a network of communicating nodes • Node: A process that performs a specific computation • Topic: A stream of messages published by a node and subscribed to by one or many nodes • Message: a data structure with typed fields, used for sending info or commands • Others: Manifests, stacks, services, bags
NIMBUS ROS System Drone I/O cluster • Runs on Linux, uses ROS system MITAscTec • MITAscTec • Collection of packages for communicating with and monitoring drones • Capabilities: • GUI interface for sending state-based commands • Integration with Vicon infrared camera system in NIMBUS lab and GPS for positioning • Remote access capabilities • Serial communication with drone using XBee module State machine Control cluster
Progress: MAVLink/ROS Node Integration • Quadcopter uses MAVLink protocol, but MITAscTec uses AscTec protocol • Need for intermediate node to translate • Integrating an open-source node • Install MAVLink/ROS package • Determine placement of new node and create launch files specifying node and topic configuration • MAVLink/ROS package: • Outbound data translated from AscTec protocol to MAVLink • Inbound data translated from MAVLink to AscTec protocol
Progress: Integration • Sending commands: • ROS command input • /a/robot_trans translates command to AscTec, /a/robot_prot handles AscTec protocol • MAVLink node translates AscTec to MAVLink and sends commands, receives data similarly
Future Plans: Integration • Test MAVLink inbound node by receiving status/heartbeat (no outbound components) • Test MAVLink outbound node with Vicon launch configuration (Relies on input from Vicon camera system) • Make alterations to MAVLink/ROS node based on differences between MAVLink and ArduCopter protocols • Build other launch configurations once MAVLink/ROS node is tested
iOS App Status • Ability to communicate with APM using UDP • When app starts, it detects heartbeat from APM and extracts APM system ID, which is then used to request a data stream that contains status information • App also accesses iPhone GPS to get current location which will be used to implement “Follow Me” function
iOS App Implementation • MAVlink is a messaging library only available in C • iOS Objective-C allows for easy integration of MAVlink library with no modificationThinking more on the stream pack(ByteBuffer) when send and receive • Using open source iOS code (AsyncUdpSocket) to implement UPD communicationEasy to sent with the port and IP address, and receive Objective-C type Data with action immediately. • Wrote the phone GPS class with the Objective-C framework.
Next Plan • Done on the display part • Available for send actually command. • Test outside with iPhone.
Android App Status • Ability to communicate using UDP and display all copter status info • Notifies user if connection is lost • All commands are implemented, but so far only Arm/Disarm is confirmed to be working • Access to Android GPS for “Follow Me” coordinates
Android App Implementation • Attempted to implement MAVlink library using JNI • Ended up writing Java classes to implement messages and commands • Only implement messages we need (about 15) • Involved working with raw byte data (ByteArray, ByteBuffer) • Wrote custom class for UDP connection that implements InputStream • Same functions as TCP InputStream (avoid refactoring code) • Utilizing Android AsyncTask • Main thread is reserved for UI • AsyncTask allows you to easily launch tasks in a new thread • One thread is dedicated to receiving and unpacking UDP communication • Each command launches an AsyncTask to send command and wait for acknowledgement from APM
Setbacks/Problems • Testing commands must be done outside • Requires GPS • Weather not cooperative • Requirement from sponsors that at least 2 people are present • How did wi-fi module break? • Replacing parts after crashes is time consuming. • Currently need to replace at least one motor
Setbacks/Problems ctd... • ArduCopter does not implement all MAVlink commands • Makes potentially easy commands much harder to implement (Takeoff) • ArduCopter has made several modifications that are not documented • Requires going through source code in detail • May add difficulty to implementation of NIMBUS node • Implemented commands on Android currently not working • Takeoff and Land commands were not working as of last week • Guided Mode vs. Auto Mode? • ArduCopter does not implement MAVlink Takeoff command, which may require sending mission script to copter and changing flight mode and throttle level