Flying with Intel® SoC FPGA · Game of Drones Processor Sensors Data Link Motor ... Flight...
Transcript of Flying with Intel® SoC FPGA · Game of Drones Processor Sensors Data Link Motor ... Flight...
Flying with Intel® SoC FPGA: Smart Drones Enabled by Open Source Platform
Zongbo Wang / CEO Aerotenna
1
Agenda
• Game of Drones
• The Path to Smart Drones
• Flying with OcPoC Open Source
Platform
• Summary
2
Game of Drones
3
4
Game of Drones
Processor
Sensors
Motor Data Link
Motor
Motor
Motor
Flight Controller
Exciting combination of electrical and aerospace engineering
and computer science, from theory to practice.
5
Current Hardware Option and Software Stack
Pixhawk with ARM* MCU
Compatible Open Hardware
(Pixhawk, APM, etc.)
Embedded OS
(RT Linux*, NuttX, etc.)
PX4 or APM Flight Code
Evolution to Smart Drones
6
7
Sensing
Processing
From Drones to Smart Drones
How to Achieve Safe and Reliable Autonomous Flying? Sensors and Sensor Fusion
• Ultrasonic, Microwave • Optical (Infrared, Vision)
Onboard Real-time Processing and AI • Remote processing is not feasible
during flight due to time sensitivity
8
Designing the Brain of the Smart Drone
Demands in Smart Flight Technical Features of SoC FPGA
Sensor Interface and Communication I/O Throughput
Sensor Signal Processing and Sensor Fusion FPGA
High-level Decision Making and Flight Control Embedded ARM* Processor
Real-time AI and Deep Learning FPGA-based Real-time Processing
9
Brain of the Smart Drone
ARM*
Processor
FPGA
SoC FPGA
Cyclone® V SoC
Basic peripheral devices for drones
Additional external sensors for smart drones
Processing System
Ardupilot on Linux*
DDR 3
IMU
Baro
GPS
Programmable Logic
PWM Coding
PWM Coding
PWM Coding
PWM Coding
PPM Decoding
RC Receiver
ESC ESC ESC ESC
UART
Rangefinder
Patten Reorganization
CSI
Camera Collision-Avoidance
Radar
2D-FFT
LVDS
10
Advantages of SoC FPGA Structure
PPM/PWM
Coding/Decoding Logic
PPM Signal from Remote Controller
PWM Signals to Motors
ARM ARM + FPGA
500+ lines of C/C++ VS 20 lines of VHDL/Verilog + 6 lines of C/C+
1000+ CPU cycles VS <5 clock cycles of pipeline delay
Free up CPU load for high-level “thinking”
FPGA
ARM* Processor
Flying with OcPoC* Open Source Platform
11
12
Introducing OcPoC* – “Octagonal Pilot on Chip”
Highlights
• First Intel® SoC FPGA-based flight controller
• Utilizes Enpirion® power modules
• >100 I/Os for sensor integration
• Video streaming and processing
• Enhanced GPS and IMU sensor packages
• APM compatible (dronecode.org)
• Open source platform (hardware, software)
Flying with OcPoC
13
14
OcPoC’s Sensor Options
Aerotenna’s Microwave Sensors OcPoC* Compatible Sensors
μLanding* Compact Altimeter
μSharp*
360° Sense-and-Avoid Radar Dual-Camera Module
Courtesy of Human+ Tech
Flight Control Platform
OcPoC
15
Aerotenna’s Microwave Sensors for OcPoC*
Compact Altimeter for All Drones • < 60 grams in weight • < 1.5 watts in power • < 4 cm in range accuracy • Up to 120 m in sensing range
360° Sense-and-Avoid Radar • < 150 grams in weight • < 2.5 watts in power • < 4 cm in range accuracy • Up to 60 m in sensing range
OcPoC Onboard Processing
16
IP Library for OcPoC*
PPM Decoding
PWM Coding
S.BUS Decoding
μLanding* Interface
μSharp* Range
Processing
Available
Interface IPs
Sensor IPs
Under Development
CSI-bus
Dual Camera Depth
Retrieving
… …
Programmable Logic
Cyclone® V SoC
Processing System
Ardupilot on Linux* PWM
Coding
PWM Coding
PWM Coding
PWM Coding
PPM Decoding
17
Dronecode in the Game of Drones
Dronecode Project Members
18
Open Source
Open Culture
Open Skies
Open to Exploring New Tech!
Dronecode.org
19
Summary
• Drones are evolving into smart drones
• Intel® SoC FPGAs fit perfectly as the brain of the smart drone
• OcPoC* is available as an open-source platform supporting drones
• OcPoC enables the development of:
- Sensor fusion
- Real-time processing
- Vision-based flying assistance
- Deep learning and environmental sensing
- and more…
Are You Ready for Smart Drones?
20
Additional Sources of Information
More web based info: www.aerotenna.com
2
0