Deterministic and high throughput data processing for CubeSats
- 1. ONBOARD HIGH RATE DATA PROCESSING USING CANOPEN PROTOCOL pablo.ghiglino@klepsydra.org www.klepsydra.com
- 2. 01: Introduction •Introduction •Use case •ROS Benchmark •Klepsydra Space •The product •The company •Conclusions
- 3. Embedded systems today Embedded autonomous systems is a fast growing market with as fast growing challenges: • Increasing demand in complexity: • Large data processing on-board (sensor data, communications with other systems, etc) • Heavy algorithms: filtering, FDIR, optimal control, vision navigation, etc. • Hardware has improved substantially: • Better on-board computers • Improved sensors (more and faster data)
- 4. Software fo embedded systems Software challenges: • Lower priority. • Lack of modern techniques. • Lack of highly skilled software engineering • This translate into: • Error prompt software • Limited functionality • Delays in delivery • Suboptimal resource utilisation
- 5. Parallel data processing • The majority of embedded software problems are related to data processing. •The financial sector has known this for a long time! This is where Klepsydra come into place
- 6. High Frequency Trading • Thousands of transactions per second. • Require access to real time data. • Delays in processing might result in millions in loses. Innovation
- 7. 02: Use case •Introduction •Use case •ROS Benchmark •Klepsydra Space •The product •The company •Conclusions
- 8. Use case: windmill inspections Client Requirements •Onboard capture, video-stream, process and record high resolution images at 20 frames per second(FPS). •Fly as fast and close as possible to the blades of the windmill and do it very robustly. •Vision-based navigation. Situation before Klepsydra •Only low resolution images and low FPS. •Error prone and slow and inaccurate navigation.
- 9. Event Loop Sensor Multiplexer Two main architectural approaches Sensor Consumer 1 Consumer 2 Sensor 2 Sensor 3 Processor Sensor 1
- 10. CAPTURE IMAGE PROCESSING FILTERING IMAGE PROCESSING (Slam, robotics) LIVE STREAM IMAGE COMPRESSION (Low priority) 1 Camera, 3 real-time uses • One big producer of data. Usually large data (image, point cloud, etc) • Several independent consumers. • Each can have different rates.
- 11. Use case: the solution With Klepsydra: High resolution camera Without Klepsydra: Low resolution camera High CPU Consumption Low CPU Consumption
- 12. Event Loop Sensor Multiplexer Two main architectural approaches Sensor Consumer 1 Consumer 2 Sensor 2 Sensor 3 Processor Sensor 1
- 13. Conclusion Now Alerion, offers the most advanced autonomous drones for windmill inspections. They produce the highest resolution images and closest captured in the market. This extremely complex application was much easier with Klepsydra solutions. They are now the official inspection provider for Siemens-Gamesa. Second biggest windmill manufacturer in the world with 15’000 windmills worldwide.
- 14. 03: ROS Benchmark •Introduction •Use case •ROS Benchmark •Klepsydra Space •The product •The company •Conclusions
- 15. The Robot Operating System • Benchmarking for ROS and ROS with Klepsydra • Test for high rate sensor data as per expected in the vision based navigation. • ROS is an asynchronous middleware.
- 16. ROS in Space • Used by Universities for simulation of Space scenarios and applications • Used by companies and organisation to simulate space robotics scenarios
- 17. Parallel data processing solutions Middleware Single Thread Spin Application Middleware Multi Thread Spin Application + concurrency contention Middleware Multi Thread Spin Application Klepsydra Single Thread Solution Multi Thread Solution Klepsydra Solution
- 18. Benchmarking SENSOR FUSION BENCHMARKING APPLICATION Server Application Producer 2 Producer 8 Client Application Consumer 2 Consumer 8 Producer 1 Consumer 1
- 19. Processing Rate Comparison SensorDataProcessingRate (Hz) 750 1'200 1'650 2'100 2'550 3'000 Sensor Data Production Rate (Hz) 750 1'313 1'875 2'438 3'000 ROS ST ROS MT Klepsydra + ROS • Pure ROS starts loosing data at 1KHz. • Klepsydra does not loose any data and remains close to ideal curve. DATA LOSS
- 20. Conclusion + =
- 21. Contents Klepsydra Space •Introduction •Use case •ROS Benchmark •Klepsydra Space •The product •The company •Conclusions
- 22. Space systems today In Space applications there is a growing need of high rate data processing: • Constellations for: • Fast communications (LeoSat), • Internet providers (OneWeb), or • IoT (LaserLight/Xenesis) 1000’s requests per second from Earth, plus routing and position coordination. etc. • Space robotics, GNC, rendezvous, planetary landing. These systems need to process data as fast as possible in order to perform the mission correctly.
- 23. Sources of data events NETWORK Application SENSORS RADIO EARTH COMMS Internal: FPGA, FDIR, OS, etc.
- 24. Benchmarking SENSOR FUSION BENCHMARKING APPLICATION Server Application Producer 2 Producer 8 Client Application Consumer 2 Consumer 8 Producer 1 Consumer 1 • Combination of: SDO, PDO and SDO block communications. • Two implementations: Klepsydra and traditional Multi- thread • Variable 5-20 parallel processes • Tested in the Zedboard..
- 25. OpenMCT Performance Monitoring • Telemetry monitoring tool developed by NASA. • Used in real missions. • Web-based, lightweight. • Klepsydra uses for performance monitoring
- 26. Benchmarking Data Processing Rate Comparison DataProcessingRate(Hz) 2300 2400 2500 2600 2700 2800 2900 3000 3100 Time (s) 0 34 66 98 126 150 184 212 238 260 290 316 354 Without Klepsydra (Multithread) With Klepsydra • At 3KHz, Klepsydra is at the ideal point • Traditional methods degrades 15% • Klepsydra has negligible variability, while traditional approach does not.
- 27. CPU Usage Comparison ProcessCPU(%) 58 63 68 73 78 83 Data Event Rate (Hz) 280 784 1288 1792 2296 2800 Multi THread Klepsydra Up to 10% less CPU Usage •Better CPU usage from the beginning. •For the same rate of data processing, the gain is much bigger (15-20%)
- 28. Data Processing Rate Comparison ProcessingRate(Hz) 700.00 1120.00 1540.00 1960.00 2380.00 2800.00 Data Event Rate (Hz) 700 1120 1540 1960 2380 2800 Multi THread Klepsydra 15% event lost without Klepsydra •At 700Hz the difference become noticeable •Klepsydra remains close to the ideal processing curve.
- 29. Variability of processingStandarDev(%) 0.00 4.60 9.20 13.80 18.40 23.00 Data Event Rate (Hz) 280 784 1288 1792 2296 2800 Multi THread Klepsydra •Klepsydra’s std dev in processing is constant •Means predictable system, which in turn means more reliability.
- 30. • Negligible latency • ‘Absorbs’ parallel processing complexity and therefore, makes software development much easier • Predictable, faster and reliable behaviour •No data losses!!! Embedded Application EVENT LOOP Conclusion
- 31. Conclusion Processing faster and more data on-board has the following benefits: • Constellations • Process more requests from Earth (= $$$) • Process more constellation position/coordination data (= larger constellations, and longer service life) • Low latency processing (= faster Earth-to-Earth communications) • Space robotics, GNC, rendezvous, planetary landing. • Low latency processing (faster response to sensor data) • Process more sensor data (higher accuracy of GNC) • Predictability (increase reliability in any conditions) Increase economical benefit and chances of mission success!!
- 32. Our Vision ESA OBDP Workshop 2019. Talk by Cornelius J. Dennehy from NASA about GN&C software: “There is a critical need to modernise on-board computing capabilities to support higher levels of GN&C Autonomy. In our view improved spaceflight computing means not only enhanced computational performance, energy efficiency, fault tolerance, but also ease of programming, affordability, reconfigurability and availability” This is exactly in line with out vision and philosophy!!!
- 33. Contents The product •Introduction •Use case •ROS Benchmark •Klepsydra Space •The product •The company •Conclusions
- 34. Technical description DevelopmentTools Matlab Integration CORE High performance module Robotics / Aerospace Add-ons Image data processing Autopilots Integration Space Add-ons Space comms connectors On-board computer optimisation Real time operating Systems Robotics Framework integration AutoCoding • ‘Miniaturisation’ of High frequency trading techniques. • General purpose library for embedded systems: Robotics, Space, Aerospace, IoT and Self- driving cars. • Platform independent.
- 35. Development tools Run it in most Linux and RTOS distributions. Integrate our product with add-ons for CAN and FPGA boards. Adoption and integration into existing applications with very small effort. Efficient and reliable autocoding code without manual modifications. Accelerate development with our enhanced processor-in-the-loop simulations. MATLAB/Simulink Direct programming
- 36. Contents The company •Introduction •Use case •ROS Benchmark •Klepsydra Space •The product •The company •Conclusions
- 37. Team •Dr. Pablo Ghiglino. Founder and CEO. •Isabel del Castillo. Admin management •Dr. Mandar Harshe. Senior C++ developer. •Franco Bugnano. Part-time software developer (space) •Dr. Francisco Sanchez. Part-time C++ developer (core and aerospace) •Javier Aldazabal. Part-time C++ developer (aerospace) •Hervé Flutto. Business development Europe.
- 38. Semi-finalist in the ESA Space Masters, Luxembourg 2017 Finalist in the Digital Energy Forum, Munich. 2017 Finalist in the BSH Venture Forum, Munich 2017 Selected by ESA to be one of the top space startups to present at IAC 2018. Awards Finalist in the NASA iTech Colorado Spring 2019
- 39. Contents The company • Summary • The product • High performance • Sensor fusion • Agile development • Competitors • The company
- 40. Next steps • We are currently doing projects with companies in the ‘NewSpace’ sector and Universities. • Klepsydra Space is available for trial: • Klepsydra + CANOpen add-on. • MATLAB add-on for beta testing.