Difference between revisions of "F17: Optimus"

Parts List & Cost

The Project bill of materials is as listed in the table below.

CAN Communication

The controllers are connected in a CAN bus at 100K baudrate. System Nodes: MASTER, MOTOR, BLE, SENSOR, GEO

DBC File

The CAN message id's transmitted and received from all the controllers are designed based on the priority of the CAN messages. The priority is as follows

Priority Level 1 - User Commands

Priority Level 2 - Sensor data

Priority Level 3 - Status Signals

Priority Level 4 - Heartbeat

Priority Level 5 - Telemetry signals to display in I/O

BU_: DBG DRIVER IO MOTOR SENSOR MASTER GEO BLE

BO_ 1 BLE_START_STOP_CMD: 1 BLE
SG_ BLE_START_STOP_CMD_start : 0|4@1+ (1,0) [0|1] "" MASTER
SG_ BLE_START_STOP_CMD_reset : 4|4@1+ (1,0) [0|1] "" MASTER

BO_ 2 MASTER_SYS_STOP_CMD: 1 MASTER
SG_ MASTER_SYS_STOP_CMD_stop : 0|8@1+ (1,0) [0|1] "" MOTOR

BO_ 212 BLE_GPS_DATA: 8 BLE
SG_ BLE_GPS_long : 0|32@1- (0.000001,0) [0|0] "" GEO
SG_ BLE_GPS_lat : 32|32@1- (0.000001,0) [0|0] "" GEO

BO_ 213 BLE_GPS_DATA_CNT: 1 BLE 
SG_ BLE_GPS_COUNT : 0|8@1+ (1,0) [0|0] "" GEO,SENSOR

BO_ 214 GEO_CURRENT_COORD: 8 GEO
SG_ GEO_CURRENT_COORD_LONG : 0|32@1- (0.000001,0) [0|0] "" MASTER,BLE
SG_ GEO_CURRENT_COORD_LAT : 32|32@1- (0.000001,0) [0|0] "" MASTER,BLE

BO_ 195 GEO_TELECOMPASS: 6 GEO
SG_ GEO_TELECOMPASS_compass : 0|12@1+ (0.1,0) [0|360.0] "" MASTER,BLE
SG_ GEO_TELECOMPASS_bearing_angle : 12|12@1+ (0.1,0) [0|360.0] "" MASTER,BLE
SG_ GEO_TELECOMPASS_distance : 24|12@1+ (0.1,0) [0|0] "" MASTER,BLE
SG_ GEO_TELECOMPASS_destination_reached : 36|1@1+ (1,0) [0|1] "" MASTER,BLE
SG_ GEO_TELECOMPASS_checkpoint_id : 37|8@1+ (1,0) [0|0] "" MASTER,BLE

BO_ 194 MASTER_TELEMETRY: 3 MASTER
SG_ MASTER_TELEMETRY_gps_mia : 0|1@1+ (1,0) [0|1] "" BLE
SG_ MASTER_TELEMETRY_sensor_mia : 1|1@1+ (1,0) [0|1] "" BLE
SG_ MASTER_TELEMETRY_sensor_heartbeat : 2|1@1+ (1,0) [0|1] "" BLE
SG_ MASTER_TELEMETRY_ble_heartbeat : 3|1@1+ (1,0) [0|1] "" BLE
SG_ MASTER_TELEMETRY_motor_heartbeat : 4|1@1+ (1,0) [0|1] "" BLE
SG_ MASTER_TELEMETRY_geo_heartbeat : 5|1@1+ (1,0) [0|1] "" BLE
SG_ MASTER_TELEMETRY_sys_status : 6|2@1+ (1,0) [0|3] "" BLE
SG_ MASTER_TELEMETRY_gps_tele_mia : 8|1@1+ (1,0) [0|1] "" BLE

BO_ 196 GEO_TELEMETRY_LOCK: 1 GEO
SG_ GEO_TELEMETRY_lock : 0|8@1+ (1,0) [0|0] "" MASTER,SENSOR,BLE

BO_ 3 SENSOR_LIDAR_OBSTACLE_INFO: 6 SENSOR
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR0 : 0|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR1 : 4|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR2 : 8|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR3 : 12|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR4 : 16|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR5 : 20|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR6 : 24|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR7 : 28|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR8 : 32|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR9 : 36|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR10 : 40|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR11 : 44|4@1+ (1,0) [0|12] "" MASTER,BLE

BO_ 4 GEO_TURNING_ANGLE: 2 GEO
SG_ GEO_TURNING_ANGLE_degree : 0|9@1- (1,0) [-180|180] "" MASTER,BLE

The CAN DBC is available at the Gitlab link below

https://gitlab.com/optimus_prime/optimus/blob/master/_can_dbc/243.dbc

CAN Bus Debugging

We used PCAN Dongle to connect to the host pc to monitor the CAN Bus traffic using BusMaster tool. The screenshot of the Bus Master log is shown below

BusMaster CAN Signal Log

Hardware & Software Architecture

Master Controller

Software Architecture Design

The Master Controller Integrates the functionality of all other controllers and it acts as the Central Control Unit of the Self Navigating car. Two of the major functionalities handled by Master Controller is Obstacle avoidance and Route Maneuvering.

The overview of Master Controller Software Architecture is as show in the figure below.

SW Architecture

As an analogy to Human driving, it receives the inputs from sensors to determine the surrounding of the Self Navigating car and take decisions based on the environment and current location of the car. The input received and output sent by the Master are as mentioned below:

Input to Master:

1. Lidar Object Detection information - To determine if there is an obstacle in the path of navigation

2. GPS and Compass Reading - To understand the Heading and Bearing angle to decide the direction of movement

3. User command from Android - To stop or Navigate to the Destination

Output from Master:

1. Motor control information - sends the target Speed and Steering direction to the Motor.

Software Implementation

The Master controller runs 2 major algorithms, Route Maneuvering and Obstacle Avoidance. The System start/stop is handled by master based on the Specific commands. The implicit requirement is that When the user selects the destination, route is calculated and the checkpoints of the route are sent from Android through bridge controller to the Geo. Once Geo Controller receives a complete set of checkpoints, the master controller starts the system based on the "Checkpoint ID". If the ID is a non-zero value, the route has started and Master controller runs the Route Maneuvering Algorithm.

The Overall control flow of master controller is shown in the below figure.

Process Flowchart

Unit Testing

Using Cgreen Unit Testing framework, the Obstacle avoidance algorithm is unit tested.The complete code for unit test is added in git project.

Ensure(test_obstacle_avoidance)
{
   //Obstacle Avoidance Algorithm
   pmaster->set_target_steer(MC::steer_right);
   mock_obstacle_detections(MC::steer_right,MC::steer_right,false,false,false,false,false,false,true);
   assert_that(pmaster->RunObstacleAvoidanceAlgo(obs_status),is_equal_to(expected_steer));
   assert_that(pmaster->get_forward(),is_equal_to(true));
   assert_that(pmaster->get_target_speed(),is_equal_to(MC::speed_slow));
}
Ensure(test_obstacle_detection)
{
   //Obstacle Detection Algorithm
   mock_CAN_Rx_Lidar_Info(2,2,6,0,2,2,4,0,2,0,5,0);
   set_expected_detection(true,false,true,false,true,false,false);
   actual_detections = psensor->RunObstacleDetectionAlgo();
   assert_that(compare_detections(actual_detections) , is_equal_to(7));
}
TestSuite* master_controller_suite()
{
   TestSuite* master_suite = create_test_suite();
   add_test(master_suite,test_obstacle_avoidance);
   add_test(master_suite,test_obstacle_detection);
   return master_suite;
}

On board debug indications

Design Challenges

The critical part in Obstacle Avoidance Algorithm is designing, 1. Obstacle detcetion 2. Obstacle avoidance. Since we get 360-degree view of obstacles, we need to group the zones into sectors and tracks to process the 360-degree detection and take decision accordingly.

Obstacle Avoidance Design

Motor Controller

Design & Implementation

The Motor Controller is responsible for the Movement and Steering action of the Car. It includes two types of motors, DC motor for movement and DC Servo motor for Steering. The Motor has an inbuilt driver called ESC (Electronic Speed Control) Circuit used the manipulate the speed and steering of the Car. It has a PWM input for both Servo Motor and DC Motor. We are using RPM sensor to take the feedback from the motor to monitor the speed.

Hardware Design

Motor Hardware Schematics

Hardware Specifications

• 1. DC Motor, Servo and ESC

This is a Traxxas Titan 380 18-turn brushed motor. The DC motor comes with the Electronic Speed Control(ESC) module. The ESC module can control both servo and DC motor using Pulse Width Modulation (PWM) control. ESC also requires an initial calibration: ESC is operated using PWM Signals. The DC motor PWM is converted in the range of -100% to 100% where -100% means "reverse with full speed", 100% means "forward with full speed" and 0 means "Stop or Neutral". Also, the servo can also be operated in a Safe manner using PWM.

As we need a locked 0 –> 180 degrees motion in certain applications like robot arm, Humanoids etc. We use these Servo motors. These are Normal motors only with a potentiometer connected to its shaft which gives us the feedback of analog value and adjusts its angle according to its given input signal.

So… How to Operate it? A servo usually requires 5V->6V As VCC. (Industrial servos requires more.) and Ground and a signal to adjust its position. The signal is a PWM waveform. For a servo, we need to provide a PWM of frequency about 50Hz-200Hz (Refer the datasheet). So the time duration of a clock cycle goes to 20ms. From this 20ms if the On time is 1ms and off time is 19ms we generally get the 0 degrees position. And when we increase the duty cycle from 1ms to 2ms the angle changes from 0–> 180 degrees. So where can it go wrong-

Servo Motor Operation

Power->> The power we provide. Generally we tend to give a higher volt batteries for our applications by pulling the voltage down through regulators to 5Vs. But we surely can-not give supply to the servo through our uC as the servo eats up a hell lot of current.

Another way to burn the servo is at certain times the supply is given directly through the battery so the uC will not blow up. But if you Give a supply say 12Volts then boom. Your servo will go on for ever.

PWM–> PWM should strictly be in the range between 1ms–> 2ms (refer datasheets) If by any mistakes the PWM goes out then boom the servo will start jittering and will heat up and heat up and will burn itself down. But this problem is easily identifiable as there is a jitter sound which if you have got enough experience with servos, you will totally notice the noise. So if the noise is there when you turn on the servo, turn it off right away and change the code ASAP.

Load— Hobby servos don’t have high load bearing capacities and as it is designed that way it always tries to adjust its angle according to signal. But here is the catch. As there is too much off load the servo cannot go further and the signal is forcing it to. So again.. heat… heat and boom. How to avoid this. Give load to the servo only in the figure of safety.

• 2. RPM Sensor

The RPM sensor above requires a specific kind of Installation. STEPS ARE:

Once the installation is done, the RPM can be read using the above magnetic RPM sensor. It gives a high pulse at every rotation of the wheel. Hence, to calculate the RPM, the output of the above sensor is fed to a gpio pin of SJOne board.

Motor Module Hardware Interface

The Hardware connections of Motor Module is shown in above Schematic. The motor receive signals through CAN bus from the Master and feedback is sent via RPM sensor to the Master as current speed of the Car. The speed sent from a RPM sensor over a CAN bus is also utilized by I/O Module and BLE module to print the values on LED display and Android App.

Software Design

The following diagram describes the flow of the software implementation for the motor driver and speed feedback mechanism.

Motor controller flowchart

Motor Module Implementation

The motor controller is operated based on the CAN messages received from the Master. The CAN messages for Drive and Steer commands are sent from the Master Controller. Motor controller converts the value received from Master (+100 to -100 for Drive Speed percent and +100 to -100 for Steer angle in the range of 1 to 180 degrees turn) into specific PWM value as required by DC motor and Servo.

• Speed Regulation:

Upon detection of uphill the pulse frequency from RPM Sensor reduces, that means car is slowing down. Hence, in that scenario, car is accelerated (increase PWM) further to maintain the required speed. Similarly in case of Downhill pulse frequency increases, which means car is speeding up. Hence, brakes (reduced PWM) are applied to compensate the increased speed.

Sensor Controller

The Sensor is for detecting and avoiding obstacles. For this purpose we have used RPLIDAR by SLAMTEC.

Introduction

LIDAR System Composition

Under this condition, the resolution will be 0.9°. And the actual scanning frequency can be freely adjusted within the 5-15hz range according to the requirements of users. The RPLIDAR A2 adopts the low cost laser triangulation measurement system developed by SLAMTEC, which makes the RPLIDAR A2 has excellent performance in all kinds of indoor environment and outdoor environment without direct sunlight exposure.

This LIDAR consists of a range scanner core and the mechanical powering part which makes the core rotate at a high speed. When it functions normally, the scanner will rotate and scan clockwise. And users can get the range scan data via the communication interface of the RPLIDAR (UART) and control the start, stop and rotating speed of the rotate motor via PWM.

A laser beam is sent out by the transmitter and the reflected laser beam is received back. Depending on the time taken to receive the beam back, the distance of the obstacle is calculated. If there is no obstacle, the beam will not be reflected back.

Hardware Implementation

Specifications of the LIDAR

The specifications of the LIDAR as mentioned in the datasheet are as follows:
Power Supply: 5V
Serial Communication interface: UART
Baud Rate for the UART: 115200
Working mode of the UART: 8N1
PWM Maximum Voltage: 5V (Typical 3.3V) PWM frequency: 25KHz
Duty Cycle of the PWM wave: 60% - 100%

Connections to the SJOne Board

The LIDAR works with a UART interface and hence has been connected to the UART3 pins of of the SJOne board i.e. P4.28 and P4.29. As the LIDAR needs a 5V supply, it is provided from the PCB (which is powered through a power bank) instead of the SJOne board which can supply only 3.3V. The connections can be seen in the figure below.

LIDAR Connections to SJOne Board

Software Implementation

Approach for obtaining the data from the LIDAR

The LIDAR senses all the obstacles around it (360 degrees upto a range of 6000cm) one degree at a time. This means that for one rotation of the LIDAR WE GET 360 values i.e. 360 angles with their corresponding obstacle information. It takes 180ms for the LIDAR to complete one 360 degree scan. Since we do not need obstacle information for each and every angle, we group a few angles together into "sectors" and consider the nearest object present in a sector as an obstacle. To identify how far an obstacle is located, the distance values are grouped into "tracks" i.e 0cm to 25cm is track 1 and 25cm to 50cm is track 2 and so on. The motor will take action depending on the track in which an obstacle is present.

LIDAR readings divided into sectors and tracks

Algorithm for interfacing LIDAR to SJOne board and obtaining the obstacle info

Step 1: Send a GET_HEALTH (0XA5 0X52) Request. If the receive times out it is a communication error.

The GET_HEALTH request and response packets

Step 2: Check if a ‘protection stop’ is happening. If it is happening then send a RESET (0XA5 0X40). Again check for ‘protection stop’ and if it it still set, send a RESET again. If ‘protection stop’ is set even after sending RESET multiple times it means there is a hardware defect. If there is no hardware defect, proceed to the next step.

The RESET request packet

Step 3: Send a START_SCAN (0XA5 0X20) request. Wait for the response header. If there is no timeout, read the measurement sample. Otherwise check HEALTH_STATUS and MOTOR_STATUS again. Send START_SCAN again.

The START_SCAN request and response packets

Step 4: Continuously read the measurement samples.The data sent from the LIDAR will contain the start bit, angle, distance and quality. The start bit is set to 1 after every single 360 degree scan. The angle and distance represent to the motor angle and the obstacle in that corresponding angle. The quality represents the strength of the reflected beam. If the quality is zero it means that there is no obstacle in that direction. This data is processed to be group the information into sectors and tracks.

The measurement response packet

Step 5: If we wish to stop the readings, send a STOP (0XA5 0X25) request. This is the end of operation.

Flowchart for Communicating with the LIDAR and receiving obstacle information

The entire flowchart for communicating with the LIDAR and receiving data from it is shown in the figure below:

Flowchart for communicating with the LIDAR and receiving obstacle information from it

Testing the data obtained from the LIDAR

To perform the initial testing of the LIDAR and to check if we are getting the correct obstacle info, we have made a setup enclosing the LIDAR on all four sides. So, by plotting the distance info given by the LIDAR in Microsoft Excel we can visualize a map of the obstacles as detected by the LIDAR. The map plotted in Excel after closing almost all four sides of the LIDAR can be shown in the figure shown below.

Data Obtained from the LIDAR plotted on an Excel sheet

CAN DBC messages sent from the Sensor Controller

The data received from the LIDAR is grouped into sectors and tracks and is sent over the CAN bus. The CAN DBC messages in the DBC file will be as follows
BO_ 3 SENSOR_LIDAR_OBSTACLE_INFO: 6 SENSOR

SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR0 : 0|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR1 : 4|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR2 : 8|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR3 : 12|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR4 : 16|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR5 : 20|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR6 : 24|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR7 : 28|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR8 : 32|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR9 : 36|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR10 : 40|4@1+ (1,0) [0|12] "" MASTER,BLE
SG_ SENSOR_LIDAR_OBSTACLE_INFO_SECTOR11 : 44|4@1+ (1,0) [0|12] "" MASTER,BLE

Android Application

Description

An Android Mobile Device Application to Navigate and trigger power to the Self Driving Car "OPTIMUS".

Optimus App serves an important role in the SDLC as it integrates the UI alongwith RC Controls like "Power", "Navigation" over Bluetooth channel with the Self navigating CAR.
The App uses RF-Comm Bluetooth Communication protocol and a BLE Transceiver to communicate with CAR and exchange several useful information over a Baud-rate of 9600bps.
Optimus mobile Platform needs to be connected with a specific Device Address based on the BLE Chip type in use.

• OPTIMUS HOME

Optimus App: OPTIMUS HOME

Features

1. BLUETOOTH

The Android mobile application includes support for the Bluetooth network stack, which allows a device to wirelessly exchange data with the HC-05 Bluetooth device.
The Android application framework provides access to the Bluetooth functionality through the Android Bluetooth APIs. These APIs let applications wirelessly connect to other Bluetooth device, enabling point-to-point wireless features.

Using the Bluetooth APIs, Android application performs the following:

• Scan for other Bluetooth device HC-05 on RC Car [00:**:91:D9:14:**]
  
• Query the local Bluetooth adapter for paired Bluetooth devices
  
• Establish RFCOMM channels
  
• Connect to other devices through service discovery
  
• Transfer data to and from other devices
  
• Manage multiple connections
  

2. MAPS

OPTIMUS App uses Google Maps for setting up the Routing Map information and to decide on the next checkpoint for the Car and the appropriate shortest route by computing the checkpoints using "Adjacency Matrix" and certain algorithms.
Google Maps are used along with other promising features to improve the navigation experience as the Route plot and Checkpoint mapping on groovy paths around campus are difficult to plan and route using Google Api(s).

• MAPS :: ANDROID - BLE COMMUNICATION JSON SCHEMA
  

The App was also upgraded to have live tracking feature of Car's location by indicating the crossed marker with YELLOW_HUE color to distinguish the original path and the traversed path by the car.
As soon as the car crosses a checkpoint marker the marker color will be updated to YELLOW from its original BLUE Color to indicate the checkpoint flag has been crossed.

Optimus App: LIVE CAR TRACKING

Optimus app uses interpolation schemes to calculate intermediate routes and to set checkpoints using Draggable Marker mechanism to set Destination and plot route path till the same.
The Json Format shown has various tags for extracting checkpoint information using Json reader and plotting the points on the Map. Features of the Json Data packet are:

• Feature Properties:
  

 * Name                   : Description of the route Start Point

 * Description [optional] : Custom Description of the route

 * LineString             : Signifies the route type eg. Line Plot

 * coordinates            : List of Lat-Long Coordinates till Next major Check point

ROUTE INTERPOLATION DATA

3. DASHBOARD

Dash Board was designed to have an at a glance View and to project a UI similar to a CAR Dashboard on the App wherein we have Compass Values, Bearing and Heading Angles, Lidar Maps to resonate the data obtained from LIDAR which also helps in debugging the features and the values being sent from respective Sensor Modules.

• OPTIMUS DASHBOARD
  

Optimus App: OPTIMUS DASHBOARD

• DASHBOARD JSON SCHEMA
  

DASHBOARD DATA

• Dashboard Information:
  

 * JSON_ID_GPS_LOCK_STAT   : Signifies the current Status of GPS LOCK on the car

 * JSON_ID_COMPASS_HEADING : Signifies current Heading Angle from COMPASS

 * JSON_ID_COMPASS_BEARING : Signifies current Bearing Angle from COMPASS

 * JSON_ID_TURNING_ANGLE   : Signifies current TurningAngle from COMPASS

 * JSON_ID_DIST_TO_DEST    : Signifies distance from Current Location to Destination or Absolute Displacement of Car relative to Destination Checkpoint

 * JSON_ID_DEST_REACHED    : Signifies whether the car has reached Destination or not!

• LIDAR Information:
  

 * JSON_ID_SENSOR_LIDAR_OBSTACLE_INFO_SEC0   : Signifies Track position of the Obstacles detected on multiple Sectors by LIDAR

For Example: Track 9, Sector 1 means Obstacle is detected at Sector 1 at 450 centimeters or 4.50 meters from the Current position of car at an angle range 20-45 degrees from LIDAR/CAR Front line of vision at that particular time instance

LiDAR detection of Track 9 Sector 1 i.e. 4.50 mts.

Android: LIDAR PLOT

Optimus App: Lidar Obstacle Detection

Android: LIDAR PLOT

Bluetooth Controller

Hardware Implementation

' Bluetooth Module Pin Configuration:’

We are using HC-05 Bluetooth module to send and receive the data from our android application.

Bluetooth Module

pin configuration

The Bridge controller is connected to the bluetooth module through the uart serial interface (Uart3) with 9600 baud rate 8-bit data and 1 stop bit.

Software Implementation

Pseudo code of Bridge controller:

1. Turn on bridge controller.

2. Initialise Bluetooth controller with Uart3 settings.

3. Initialise CAN-BUS with 100 kbps speed.

4. Handle Incoming IO messages it received from the Geo and the Sensor over CAN Bus.

5. Send the received CAN message to the Android over Bluetooth each second.

6. Send the heartbeat message every second to the Master controller.

7. Read Bluetooth message it received from the Android app.

8. Forward the Android message to GEO controller if it received checkpoints otherwise forward it to Master.

Process Flowchart

DBC format for messages sent from Bluetooth controller :

BO_ 1 BLE_START_STOP_CMD: 1 BLE
SG_ BLE_START_STOP_CMD_start : 0|4@1+ (1,0) [0|1] "" MASTER
SG_ BLE_START_STOP_CMD_reset : 4|4@1+ (1,0) [0|1] "" MASTER

BO_ 38 BLE_HEARTBEAT: 1 BLE
SG_ BLE_HEARTBEAT_signal : 0|8@1+ (1,0) [0|255] "" MASTER

BO_ 212 BLE_GPS_DATA: 8 BLE
SG_ BLE_GPS_long : 0|32@1- (0.000001,0) [0|0] "" GEO
SG_ BLE_GPS_lat : 32|32@1- (0.000001,0) [0|0] "" GEO

BO_ 213 BLE_GPS_DATA_CNT: 1 BLE 
SG_ BLE_GPS_COUNT : 0|8@1+ (1,0) [0|0] "" GEO,SENSOR

Geographical Controller

Introduction

GPS and Compass Module:

GPS:

GPS

GPS is a global navigation satellite system that provides geo location and time information to a GPS receiver anywhere on or near the Earth where there is an unobstructed line of sight to four or more GPS satellites.

Compass:

Compass

A compass is an instrument used for navigation and orientation that shows direction relative to the geographic cardinal directions (or points). Usually, a diagram called a compass rose shows the directions north, south, east, and west on the compass face as abbreviated initials. When the compass is used, the rose can be aligned with the corresponding geographic directions; for example, the "N" mark on the rose really points northward. Compasses often display markings for angles in degrees in addition to (or sometimes instead of) the rose. North corresponds to 0°, and the angles increase clockwise, so east is 90° degrees, south is 180°, and west is 270°. These numbers allow the compass to show azimuths or bearings, which are commonly stated in this notation. We are using DJI’s NAZA GPS/COMPASS to get the GPS coordinates and Heading angle. The diagram of the module is as follows:

GPS and Compass Module

Message Structure:

• GPS':

The 0x10 message contains GPS data. The message structure is as follows:

GPS Data

• Compass:

The 0x20 message contains compass data. The structure of the message is as follows:

Compass Data

• Calibration':

Why calibrate the compass?

Ferromagnetic substances placed on multi-rotor or around its working environment affect the reading of earth’s magnetic field for the digital compass. It also reduces the accuracy of the multi-rotor control, or even reads an incorrect heading. Calibration will eliminate such influences, and ensure MC system performs well in a non-ideal magnetic environment.

When to do it?

• The first time you install Naza compass.

• When the multi-rotor mechanical setup has changed.

• If the GPS/Compass module is re-positioned.

• If electronic devices are added/removed/re-positioned.

Hardware Connection

The Pin Configuration is as follows:

Block Diagram

Software Design

Algorithm: Distance calculation:

We are using the ‘haversine’ formula to calculate the great-circle distance between two points – that is, the shortest distance over the earth’s surface

Bearing Angle calculation:

The bearing of a point is the number of degrees in the angle measured in a clockwise direction from the north line to the line joining the centre of the compass with the point. A bearing is used to represent the direction of one point relative to another point. The bearing angle is calculated by using the following formula:

Angle Information

DBC Messages

Flowchart

Package Design

PCB Design

PCB Complete Schematic for All 5 Control Interfaces

PCB Complete Board design for All 5 Control Interfaces

3D Printed Sensor Mounts

We designed 3D printing Models for holding the Sensor LiDAR and GPS using OpenScad Software.

LiDAR Mount

GPS Mount

Git Project Management

The Gitlab project is managed using working on different branches for different controllers and restricting access to all users to merge the branch to master branch. To get easy notification of all git activity, we created a webhook for git notifications in CMPE243 Slack Channel. The useful features of git such as Issues List, Milestone tracks are used for easy management

GitLab WebHooks

The project git repository is below.

https://gitlab.com/optimus_prime/optimus/tree/master

Technical Challenges

Motor Technical Challenges

1) ESC Calibration
We messed up the calibration on the ESC.
XL 5 had a long press option to calibrate the ESC, where the ESC shall:
a) After long press, glow green and start taking PWM signals for neutral (1.5).
b) Glow green once again where we shall feed in PWM signals for Forward (2ms).
b) Glow green twice again where we shall feed in PWM signals for Reverse (1ms)."
-We wrote code to calibrate using EXT-INT (EINT3) over P0.1 - switch to calibrate the ESC this way!

2) ESC Reverse
The ESC was not activating reverse if we directly - as in the datasheet (no formal datasheet - only XL 5 forums - talked about 1ms pulse width at 50Hz for reverse).
We figured out that Reverse is actually 3 steps:
a) goNeutral()
b) goReverse()
c) goNeutral()
d) goReverse()

3) RPM Sensor Installation:
After following the steps to install RPM sensor (as steps above), the RPM sensor was not detecting the Rotation (magnet) of the wheel.
The reason for that was Machine steeled pinion gear and slipper clutch. The Machine steeled pinion gear and slipper clutch that came with the RC car was big. That increased the distance between Magnet and RPM sensor. That's why we were not able to detect RPM of wheel.
We even checked the activity using Digital Oscilloscope.
Then we changed the smaller Machine steeled pinion gear and slipper clutch and reinstalled the RPM sensor and it worked.

Android Issues Undergone

• MAPS: Plotting Routes and Offline Check Points Calculation
  

With our initial implementation using Google Android API we were able to route maps but sooner during testing of the route navigation we faced a couple of issues as follows:

1. For Straight Line Routes, often the intermediate checkpoints were not received, as according to Google Api's checkpoints are only generated at the intersections where the route bends.
2. Due to the aforesaid drawback on straight routes it was hard to navigate and interpolation was required to make sure the GEO has enough checkpoints to redefine the heading angle before the car goes too far from its destined straight route path.
3. Google Route's are calculated from any point on the ground to the nearest offset point on the pre-drawn custom Google poly-line path, as a result the route from certain locations ended up to be on the sharp edge routes rather than smooth curves which also led to little longer routes and our car ended up in side walks or side bushes while correcting its course to follow the main route.

Optimus App: Navigation and Route Selection

• Application Compatibility
  

During Implementation one of the issues faced were the security features of Android applications and permissions to use Geo Locations and App Storage.
Every time after fresh app Installation the permissions had to be revisited and enabled for the app to access them, something which still can be upgraded further.

Testing and Procedures to Overcome Challenges

MAP DEBUGGING & ROUTE CALCULATION

For overcoming the problem of placing routes and calculating the shortest path we decided to interpolate routes in the university premises.
Steps involved:

• Draw polylines routes over saved checkpoint coordinates by reading and parsing a json file at the app level to get the next checkpoint coordinates.
  
• Use Dijkstra's Algorithm to calculate shortest path between those routes.
  
• For longer routes two approaches could be taken to calculate the intermediate checkpoints:
  
  • a. Straightline Approach
    
  • b. Geodesy Engineering Approach.
    

Source:wikipedia.org::Dijkstra's algorithm

Geodesy approach is complex and can be implemented using 'Haversine' technique to calculate the intermediate points between two points along the geographic surface of the earth but since the distances for the demo were not so long enough that can be significantly impacted by the curvature we decided to go with the primary approach.

We used Vincenty formula to compute the interpolated points between two checkpoints when the distance between the two exceeded ~(10±5)meters the algorithm will interpolate the route to give intermediate checkpoints which will be marked on the map using BLUE Markers.

For easy user view we added Hybrid TYPE MAP on the app so that user can have a 3D feel of the route.

MARKERS

• We also added colored Markers for denoting following:
  

  >>> START/STOP : Custom Markers

  >>> CAR LOCATION : Yellow Markers

  >>> INTERMEDIATE CHECKPOINTS : HUE_BLUE Markers

Optimus App: Map Markers

Sensor Controller

1. LIDAR is not able to detect black colored objects sometimes as the light from the LASER is completely absorbed by black and nothing is reflected back.

LIDAR doesn't detect black objects

2. LIDAR object detection will be the plane where it is mounted. So, if the object height is less than the height the LIDAR is mounted then the object will not be detected.

LIDAR doesn't detect objects lower than it's height

3. If there is very high ramp then ramp will also come in the plane of the LIDAR and it will be considered as an obstacle.

4. LIDAR's Exposure to direct sunlight will cause noise creation in the obstacle detection.

Geo Technical Challenges

The first and the major issue we faced with the GEO module was selecting the proper hardware for GPS and Compass. We tried with Sparkfun, Adafruit and Ublox GPS modules. We observed a lot of time taken by the GPS to get a lock and also the error was high. Then we switched to DJI Naza GPS and we found that it was pretty accurate and the lock up time was hardly a minute. The software issue which we faced with Naza GPS was that it did not have a proper software documentation. We tried to understand the message packets and went through the forums to understand the message layout. After this we were able to integrate the module successfully.

The Naza gps module comes with a in-built compass and it simplified our setup as we did not have to integrate two separate modules.

We faced one more hardware issue once the Rx pin of the gps module was accidentaly connected to the ground pin then the gps started to draw a lot of current. So to avoid this kind of mistakes we integrated fuse with the gps so even if extra current is drawn the fuse will take care that this does not hamper the entire system.

Also, the car was going to the edges even if the path was towards the middle of the road as per google maps. So after developing an app to map the checkpoints we found that the path is actually inside the buildings. So, we had to find a different solution to solve this problem. Afterwards, we created a database of all the routes in campus and then processed the route through the android app.

GPS Route

Project Videos

https://youtu.be/lzW2ASbNfYo

Conclusion

As a team we were able to achieve the set of goals and requirements within the required time frame. Over the course of this project, we learnt cutting edge industry standards and techniques such as:

• Team Work: Working in a team with so many people gave us a real sense of what happens in the industry when a large number of people work together.
• GIT: Our source code versioning, code review sessions and test management was using GIT.
• CAN: A simple and robust broadcast bus which works with a pair of differential signals. We were able to use the CAN bus to interconnect five LPC1758 micro controllers powered by FreeRTOS.

• Accountability: Dealing with both software and hardware is not an easy task and nothing can be taken for granted, especially the hardware.
• Hardware issues:
  • Power Issues: Initially we were using a single port from the Power bank power up everything (all the boards) including the LIDAR. This caused the LIDAR to stop working due to insufficient current. It took a while for us to figure this out.
  • GPS: Calibrating the GPS and getting accurate data from the GPS was a challenging task.
  • Android Application: Using google maps to obtain checkpoints did not workout as google maps was giving a single checkpoint. So we created a database of checkpoints for navigating the car across SJSU campus.
  • Debugging: Connecting the PCAN dongle to the car and moving around with it is a difficult way to debug. Hence we created a dashboard on the android application to view all the useful information on the tab without any hassles.

To the teams that are designing their car:

• If using a LIDAR for obstacle avoidance make sure to test it in all lighting conditions.
• It is better to have PCB instead of soldering everything on a wire-wrapping board.
• Start with the implementation for the Geo module early.

Project Source Code

The source code is available in the below github link

https://gitlab.com/optimus_prime/optimus

References

• ESC Calibration
• ESC PWM information
• ESC XL-5 PWM information
• GEO Bearing information
• LIDAR SDK which helped us in coding for the LIDAR
• LIDAR Data Sheet
• LIDAR User Manual

Acknowledgement

We are thankful for the guidance and support by

Professor

• Preetpal Kang

ISA

• Prashant Aithal

• Saurabh Ravindra Deshmukh

• Purvil Kamdar

• Shruthi Narayan

• Parth Pachchigar

• Abhishek Singh

For 3D printing

• Our sincere thanks to Marvin Flores <marvin.flores@sjsu.edu> for printing our 3D print models.

For Sponsoring R/C car