• WSDC 6

    Przestrzeń from Facebook

    3 pretentions: 1. Can a robot race like a pro? State of the art of high-velocity vehicle control. By Adam Gotlib 2. From zero to simple self-driving car using DonkeyCar. By Michał Sochoń 3. Training a neural network for driving an autonomous RC car. By Maciej Dziubiński -------- 2: Michał Sochoń is a Senior Systems and Network Engineer, helping within a team and organization in overall automation, especially in CI/CD integration, managing and helping from moving from testing to production systems idependently if it is on-premise, hybrid or pure cloud setup. He always wanted to have a RC car and also was recently getting interested with ML/AI, but never had time to even start it. By coincidence he discovered DonkeyCar project, which integrates both technologies, but in the pricing range of a hobby and not a full-time job. In his presentation we will be able to get more details how to start with self-driving car in 1:16 scale from complete zero. -------- 3: I will present a RC-based model of an autonomous car that can drive around a pond next to my house. The main sensor is an Intel D435i depth camera (IMU included) and the main computational unit is the Jetson TX1 which has a GPGPU which I utilized for faster inference. The car is controlled by a tiny convnet (trained on low-res depth images) that yields predictions in 3.5ms (on average). The presentation will be in a form of Q&A where I'll try to answer questions like: how to build a car like this? where to look for resources / inspirations? what limitations should one be aware of before starting? what else can one do with a simple model like this? and, hopefully, other questions from the participants of the talk. The presentation is an extension / discussion of a blog post published on Medium: https://medium.com/asap-report/training-a-neural-network-for-driving-an-autonomous-rc-car-3906db91f3e ======= === Important info === Building closes at 6pm, but you will still be able to enter, there will be more information on window.

    5
  • WSDC 5

    Startberry

    WSDC 5 English slides + PL or EN presentations related to self-driving. 18:00 - Google Street View image of a house predicts car accident risk of its resident; Kinga Kita-Wojciechowska 18:40 - RC to Full Autonomy - The Journey; Paweł Misiewicz 19:20 - Modelling the top-down view from side cameras in the CARLA simulator; Maciej Dziubiński 20:00 Pizza & networking =========================== Kinga Kita-Wojciechowska: "Google Street View image of a house predicts car accident risk of its resident" Road traffic injuries are a leading cause of death worldwide. Proper estimation of car accident risk is critical for appropriate allocation of resources in healthcare, insurance, civil engineering, and other industries. We show how images of houses are predictive of car accidents. We analyze 20,000 addresses of insurance company clients, collect a corresponding house image using Google Street View, and annotate house features such as age, type, and condition. We find that this information substantially improves car accident risk prediction compared to the state-of-the-art risk model of the insurance company and could be used for price discrimination. From this perspective, public availability of house images raises legal and social concerns, as they can be a proxy of ethnicity, religion and other sensitive data. https://arxiv.org/abs/1904.05270 Kinga is a researcher at the University of Warsaw looking for innovation in insurance pricing through application of AI. Prior to that she was working 10 years in motor insurance pricing, most of the time at AXA Group, where she held various managerial and expert roles in Poland, France, Spain, Korea, Japan and China. Holds a double master degree in mathematics and economics from the University of Warsaw. =========================== Paweł Misiewicz: "RC to Full Autonomy - The Journey" In this talk, I will summarize my journey of developing Fully Autonomous RC Car capable of driving in small, well mapped areas - the project that I started almost one year ago. The work is by no means finished, but I hope to share some useful experience with everyone who wants to move from simulators to the real world. I will present project goals, my findings and challenges encountered during preparation of the car, compute platform, data aquisition and processing. Paweł Misiewicz is an A.I. and Autonomous Vehicles enthusiast who spends his free time on R&D in those fields. He obtained his master's degree from Faculty of Mathematics, Informatics, and Mechanics of the University of Warsaw in the field of Distributed Systems. For almost 20 years he has been taking part in various projects for largest Telcos and Banks in Poland as a technical lead. For the last 10 years he has been specializing in Big Data and Data Warehousing projects. =========================== Maciej Dziubiński: "Modelling the top-down view from side cameras in the CARLA simulator" I will go through the details of building a model for predicting the top-down view from a camera hanging 100m above a car, based on input from four side cameras mounted on the car. The presentation will expand on what was already described in the blog post: https://link.medium.com/RGSyqVv2EW but I'll focus more on the topic "what didn't work and why?" Bio: Maciek works as a machine learning engineer at Nethone -- a company specializing in detecting fraudulent credit card transactions. But after work he devotes his time to autonomous vehicles and is primarily interested in applying machine learning in that field.

    5
  • Warsaw Self-Driving Cars #WSDC 4

    Startberry

    In English We will have 2 presentations and pizza: 18:00 AI-based traffic management system for connected self-driving cars, Paweł Gora 19:00 Datasets and how to create one?, Karol Majek, Deep Drive PL 19:40 Pizza & networking =========================================== Paweł Gora: AI-based traffic management system for connected self-driving cars In this talk, I will present the idea for a traffic management system, which uses traffic simulations to evaluate the qualities of the traffic signal settings, machine learning algorithms to approximate outcomes of simulations and evolutionary algorithms or reinforcement learning to find best traffic control strategies. The idea is under development within the TensorCell project, which I lead, and has been already presented in a few scientific papers and on the best conferences dedicated to transportation and artificial intelligence. The solution may be applied for traffic including only conventional cars, but I will explain why it may be especially powerful and beneficial in the era of connected and autonomous vehicles. BIO: Paweł Gora is a researcher and Ph.D. candidate at the Faculty of Mathematics, Informatics, and Mechanics of the University of Warsaw. His research concerns mostly modeling and optimization of complex processes, such as vehicular traffic in cities. He developed software for realistic, large-scale traffic simulations, Traffic Simulation Framework, and is leading a research group TensorCell working on approximating outcomes of traffic simulations using machine learning algorithms (e.g., neural networks, LigthGBM), which may find applications in transport planning and real-time traffic management systems. He received the "LIDER ITS" award in 2015 and 2017 for the best R&D work in the intelligent transportation systems domain in Poland. He is also working on traffic models including connected and autonomous vehicles and is a representative of Poland in the COST Action "Wider Impacts and Scenario Evaluation of Connected and Autonomous Transport". As an AI and quantum computing enthusiast, he is one of the co-organizers of Warsaw.ai meetup addressed to AI experts and "Quantum AI" Facebook group, as well as "Warsaw Quantum Computing Group". In 2017, he was recognized by MIT Technology Review as 1 of 10 "Top Polish Talents Under 35", and placed on a "New Europe 2017" list of 100 emerging technology stars in Eastern Europe. In the past, he worked as a software engineer and research intern at Microsoft, Google, CERN, and IBM Research. =========================================== Karol Majek Datasets and how to create one? In this talk, you will learn about currently available datasets which can be used to train neural networks or benchmark your solutions. You will learn what you can expect from public datasets and in the second part of this talk, you will learn what you need to know before you create your own dataset. You can expect knowledge and best practices on how to work with sensors and data. Bonus - release of new dataset! BIO: Karol Majek is working on Deep Drive PL, a blog related to Self-Driving, Deep Learning, but during the day he is working at NASK in mobile robotics department. He wants to share his knowledge and help people learn, so he created the Warsaw Self-Driving Cars Meetup, a place to meet people interested in Self-Driving Cars, exchange experience, learn, find a new job or people who want to work in the field. Previously working as a Mentor at Udacity in Self-Driving Car Engineer Nanodegree.

    9
  • Warsaw Self-Driving Cars #WSDC3

    Startberry

    This meetup will be held in English (!) We will have 3 presentations and pizza: 18:10 Simulating self-driving trucks: lessons learned - Adam Dąbrowski, ROBOCAR Technologies 19:00 Review of Udacity Self-Driving Car Engineer Nanodegree - Karolina Zoń 19:30 Developing a testing environment in the CARLA simulator - Maciej Dziubiński, OPIUM.sh 20:10 Pizza ========================== Simulating self-driving trucks: lessons learned At ROBOCAR Technologies we take on challenges of self-driving in less common environments. Instead of urban roads and highways, we focus on confined and semi-controlled areas such as harbor and mine roads. Simulation is key in the process of developing a system that will actually end up on a real truck. Aside from getting to know our experiences and seeing some of our results, you can gain insight into prototyping and validation of self-driving algorithms with the use of tools such as ROS, Gazebo, and Unity. The talk will also include some deep learning solutions based on LIDAR data. ROBOCAR Technologies: ROBOCAR is a software company founded by engineers and researchers that worked for last years with automotive applications, machine learning, and human-robotics interactions technologies. Our teams collaborate with leading technology suppliers and OEMs to develop robust and safe applications for automated vehicles. Website: http://www.robocartech.com/ Twitter: https://twitter.com/robocartech Adam Dąbrowski: A software engineer with more than 10 years of experience in C++, Adam graduated from the University of Warsaw (MIMUW) with MSc in Information Technology. Taking a leap of faith he quit working for corporations on semi interesting projects and moved into robotics almost five years ago. Adam now leads robotic projects at ROBOCAR, works for a research institute PIAP and also serves as an independent expert evaluator for European Commission. ========================= Review of Udacity Self-Driving Car Engineer Nanodegree SDCND from Udacity seems to be the most extensive course on various topics on Self-Driving Cars that you can find, that covers the field from different aspects - from computer vision, through sensor fusion, path planning, motion control, up to integration of all systems in a final project being later deployed on a real self-driving car. Seems very exciting if you are interested in SDC and want to get in-depth knowledge about the technical stack. But it is also quite expensive course. What can you actually learn, what projects will you implement and is it worth its price? On this presentation I will try to answer by sharing my own experience from taking on that course. Karolina Zoń: Backend developer in the ecommerce-based company, working with Java, PHP and Golang. Co-organizer of Google Developer Group Silesia and Women Techmakers Silesia. Graduated from Adam Mickiewicz University in Poznań in Computer Science with specialization in Intelligent Systems. Fascinated about robotics, machine learning, and computer vision, especially when applied in the automotive industry. After graduation deepening knowledge on those topics through side projects and courses, including Self-Driving Car Engineer Nanodegree from Udacity.

    6
  • Warsaw Self-Driving Cars #WSDC2

    Startberry

    Event in Polish. Drugi meetup tej grupy! Chcemy połączyć ludzi zainteresowanych samochodami autonomicznymi. Będziemy rozmawiać o zastosowaniu sieci neuronowych w samochodach autonomicznych, obecnych osiągnięciach, a także jak zrobić własny samochód autonomiczny 1:10 albo 1:1. Chcesz prezentować? Masz ciekawe doświadczenia? Zapraszamy! Zgłoś się do Karola. ======== Będzie prezentacja samochodu F1/10! ======== W planie 3 prezentacje: 18:10 - 18:40 Self-Racing Cars,[masked] - o zawodach samochodów autonomicznych na torze wyścigowym, Karol Majek 18:40 - 19:10 Doświadczenia z zawodów F1/10, Łukasz Sztyber / Karol Majek 19:10 - 19:40 First steps in traffic analysis - vehicle detection and simple tracking, Michał Drzał 19:40 - 20:10 System autonomicznej jazdy bazujący na mapowaniu i lokalizacji, Adam Gotlib 20:10 Pizza + dyskusja Serdecznie zapraszamy na kolejne spotkanie!

    3
  • Warsaw Self-Driving Cars #WSDC1

    Startberry

    Event in Polish. Pierwszy meetup tej grupy! Nie wiesz czego możesz się spodziewać? Chcemy połączyć ludzi zainteresowanych samochodami autonomicznymi. Będziemy rozmawiać o zastosowaniu sieci neuronowych w samochodach autonomicznych, obecnych osiągnięciach, a także jak zrobić własny samochód autonomiczny 1:10 albo 1:1. Chcesz prezentować? Masz ciekawe doświadczenia? Zapraszamy! Zgłoś się do organizatora. Na pierwszym spotkaniu zrobimy wstęp i 3 prezentacje [razem 4]: 18:10 - 18:30 Wstęp, prezentacja otwierająca 18:30 - 18:50 Jak zrobić własny samochód autonomiczny w skali 1:10 - Calvin F1/10 Team 19:00 - 19:20 KNR PW Meil [TBC] 20:00 - 20:20 HD Maps - Janusz Będkowski, Tom Tom Planujemy spotkanie przy pizzy po ostatniej prezentacji! Serdecznie zapraszamy na pierwsze spotkanie!

    2