1 00:00:10,440 --> 00:00:15,390 Welcome to this course of events and filtering for sensor fusion, and in this course, we're going 2 00:00:15,390 --> 00:00:19,650 to be covering the theory and a practical implementation for these filtering techniques. 3 00:00:21,430 --> 00:00:28,240 So why focus on common filtering and sense of fusion, data fusion is an amazing tool that is used in 4 00:00:28,240 --> 00:00:33,730 pretty much every piece of modern technology that involves any kind of sensing measurement or automation 5 00:00:33,730 --> 00:00:34,390 or control. 6 00:00:34,870 --> 00:00:38,770 The Common Filter is one of the most widely used methods for data fusion. 7 00:00:39,070 --> 00:00:44,440 By understanding this process, you will be more easily able to understand more complicated methods 8 00:00:44,440 --> 00:00:45,040 later on. 9 00:00:45,820 --> 00:00:52,750 Sensor fusion is one of the key uses of the common filter and is extensively used in unmanned vehicles 10 00:00:52,750 --> 00:00:56,410 and self-driving cars, which are both a very large, evolving fields. 11 00:00:57,690 --> 00:01:02,190 And then lastly, of course, so you don't have to waste your time trying to solve whatever problems 12 00:01:02,190 --> 00:01:07,080 that would be easily avoided with the knowledge inside this course, this course should help you become 13 00:01:07,080 --> 00:01:09,840 a subject matter expert and all things come and filtering. 14 00:01:11,710 --> 00:01:17,410 So this course is for many different people, is for university students or independent learners, is 15 00:01:17,410 --> 00:01:23,710 for aspiring robotics or self-driving cars, engineers or enthusiasts, is for working engineers and 16 00:01:23,710 --> 00:01:29,560 scientists who want to brush up on the knowledge or learn new skill sets, is what engineering professionals 17 00:01:29,590 --> 00:01:36,130 who want to understand the math and skills relating to common filtering and sense of vision is for software 18 00:01:36,130 --> 00:01:42,400 developers who wish to understand the basic concepts behind that diffusion to aid in the implementation 19 00:01:42,400 --> 00:01:45,250 or support of developing data fusion code bases. 20 00:01:45,940 --> 00:01:50,560 And as for anyone who's already proficient with the math in theory and actually want to learn how to 21 00:01:50,560 --> 00:01:51,950 implement the common filters. 22 00:01:52,840 --> 00:01:55,000 So quite a bit is covered in this course. 23 00:01:55,000 --> 00:01:59,680 We're going to be looking at a quick background of basic probability and system theory. 24 00:02:00,130 --> 00:02:02,580 We're going to be talking about the linear common filter. 25 00:02:02,590 --> 00:02:07,150 So the first type of filter that we're going to learn, then we're going to be extending that into nonlinear 26 00:02:07,150 --> 00:02:11,860 systems for the extend the common filter and then for the unscented common filter. 27 00:02:12,580 --> 00:02:18,040 We will also cover some more advanced topics for sense of fusion, such as Falke detection and sensor 28 00:02:18,040 --> 00:02:19,240 error modeling. 29 00:02:20,230 --> 00:02:26,620 We will look at the C++ implementation of a simulation for a self-driving car, in a sense, a fusion 30 00:02:26,620 --> 00:02:26,920 problem. 31 00:02:26,920 --> 00:02:31,990 So fusing different senses that might be aboard a self-driving car and using these types of filters 32 00:02:31,990 --> 00:02:38,200 to come up with the best estimate that represents the state of the system, there are some assumed knowledge 33 00:02:38,200 --> 00:02:39,010 for this course. 34 00:02:39,020 --> 00:02:42,670 So this is a more advanced course than a basic beginner's course. 35 00:02:43,330 --> 00:02:45,630 This course is part of the event series. 36 00:02:45,630 --> 00:02:49,840 So some of the background knowledge in the course is already being assumed knowledge. 37 00:02:50,020 --> 00:02:53,660 And these topics are sort of basic algebra and calculus. 38 00:02:53,680 --> 00:02:58,240 So looking at functions, derivatives integrals is looking at the linear algebra. 39 00:02:58,260 --> 00:03:05,350 So matrix and vector operations is also assume that you have some basic probability or knowledge, basically 40 00:03:05,350 --> 00:03:07,000 squares, estimation, knowledge. 41 00:03:08,060 --> 00:03:10,620 And basic C++ programming knowledge. 42 00:03:11,120 --> 00:03:14,630 So if you know all this, then you'll be in good stead for the course. 43 00:03:14,930 --> 00:03:19,880 If you're lacking in any of these areas, I would recommend brushing up or taking another course to 44 00:03:19,880 --> 00:03:21,610 get you up to speed with these areas. 45 00:03:24,010 --> 00:03:29,610 So by the end of this course, you have another outcomes, you will know how to use the linear comma 46 00:03:29,620 --> 00:03:32,790 filter to solve linear optimal estimation problems. 47 00:03:33,640 --> 00:03:38,150 You know how to use the extended Campfield to solve nonlinear estimation problems. 48 00:03:38,620 --> 00:03:44,200 You also know how to use the uncynical material for the same types of nonlinear estimation problems. 49 00:03:45,430 --> 00:03:50,800 You understand the differences between the different filter types, the advantages and disadvantages 50 00:03:50,800 --> 00:03:51,340 of each. 51 00:03:51,580 --> 00:03:57,340 So when you should use one compared to when you should use the other, you also learn how to use measurements 52 00:03:57,340 --> 00:04:04,030 of Modibo sensors or running at different update rates at the same time so the filter can operate from 53 00:04:04,030 --> 00:04:04,440 Modibo. 54 00:04:04,450 --> 00:04:05,540 Different data sources. 55 00:04:06,490 --> 00:04:10,270 You also learn how to tune the common filter for the best performance. 56 00:04:10,570 --> 00:04:14,160 So this is an important step inside the implementation of a common filter. 57 00:04:14,170 --> 00:04:16,930 It has to be tuned for the problem that you want to solve. 58 00:04:19,600 --> 00:04:25,420 You also know how to correctly initialize the common Philidor for robust operation, which is very important, 59 00:04:25,670 --> 00:04:30,040 you know, how to model sensor errors inside the computer and then correct for them. 60 00:04:30,670 --> 00:04:36,250 You will also know how to use fault detection to remove bad sensor measurements or erroneous measurements. 61 00:04:37,330 --> 00:04:43,630 And you know how to implement the three different common field of variance in C++, and lastly, you'll 62 00:04:43,630 --> 00:04:48,910 know how to implement the linear common filter, the extended filter and unscented common filter in 63 00:04:48,910 --> 00:04:54,460 C++ for an autonomous self-driving car problem, fusing multiple different types of sensors.