Login

POINT CLOUDS CONQUEROR: Master the Industry Reserved Secrets to LiDARs & 3D Perception

Dive into the world of LiDARs & Point Clouds, learn 3D Machine Learning, and build advanced Point Clouds Processing projects.

Ready to Buy

I will give you access to the Industry's Most Advanced and Reserved LiDAR Knowledge, and help you build 3D Perception Skills to become a...

POINT CLOUD CONQUEROR!

Dear Engineer,
If you'd like to build high paying "field level" LiDAR skills and expertise, that only industry professionals get access to, then this page will show you how.

Here's the story:

Over the last few years, I had the opportunity to work with CTOs, Engineers, and VCs working in the most technically advanced LiDAR companies on Earth...

It started when I first built my "Point Clouds Fast Course", which is a retired v1 of this course, and allowed me to chat with companies like Waymo, and it continued in the recent years, when launching my LiDAR related products, or when building technical content for LiDAR companies, and of course...

When travelling and meeting LiDAR experts 1-1 directly!

People like:
  • Aeva's Team of Engineers, some of team coming from Apple Titan Project, and Waymo's Firefly project, who explained me 1-1 the secrets behind FMCW LiDAR technology
  • LiDAR Researchers like Holger Caesar, the inventor of the PointPillars algorithm; who I recently met, and who taught me many things about research in the LiDAR space
  • Countless field experts and engineers, who taught me about skills needed in the LiDAR space, such as point clouds feature extraction, registration, surface reconstruction, and a few more...
  • And virtually every LiDAR company that agreed to meet me, including Ouster, Velodyne, Outsight, Aeye, mobileye, Innoviz, bitsensing, and many more. 
Each of these person has spent time teaching me a key aspect of the LiDAR world, such as the different types of LiDAR systems, how to "read" a LiDAR's technical product page, why some LiDARs are good and others aren't, how FMCW technology works, how to modify a LiDAR wave based on the weather conditions, and more...

As a result...

I have developed what I call "field knowledge" on LiDARs

By this, I mean knowledge that can only be acquired when working directly in the field. I had some of it through my experience as a self-driving car engineer, but I didn't have the full picture until very recently.

What is this field knoweldge?
It involves the answers to questions like:
  • Is X company really building a good product? Or is it overhyped?
  • How do LiDARs really work? What about optics? photons? waves? modulation? frequency?
  • What trends are happening in LiDARs? Where to invest time or money on?
And basically, all of this knowledge that you will NEVER find in an online course, and that you can't know unless you're in the field...

Since my skills in LiDARs have 10x in the past 3 years, I decided to EXECUTE my original Point Clouds Course, combine the technical Point Clouds & 3D Machine Learning skills engineers need, with the newly acquired "Field Knowledge", and create...

A new LiDAR course that will 1) Teach you hardcore LiDAR skills only the field engineers know... and 2) Teach you advanced Point Clouds Processing skills needed to build a career in the field.

So please meet my new creation...

So meet the new version of the course...

POINT CLOUDS CONQUEROR

In this course, you will learn advanced LiDAR knowledge, and combine that with Point Clouds processing skills, to build one of the most in-demand profile a company can dream of hiring...

Let's see the program:

MODULE 1

Introduction to LiDARs

In module 1, you'll build a deep understanding of Light Detection And Ranging technology. Not just "LiDARs are good with distances". Not just "they measure the time a wave takes to come back" — but a deep understanding of the most common types of LiDARs, and how they work.

What you'll learn:

  • A 3 minute introduction to the pure magic of LiDARs, explained to any beginner

  • How to read a Point Cloud File, and the two types of LiDAR file encoding that exist (and which one LiDAR engineers use)

  • The 3 Types of Scanning Systems used by LiDARs, and why Solid-State LiDARs may be replaced (I'll also share a former colleague's opinion on Solid-State technology, and explain what the field engineers seeks for in a LiDAR)

  • The major flaw shared by 99% of LiDARs that could make the entire industry collapse when scaling (disclaimer: while the LiDAR industry is solid today, it may not be as solid if hundreds of LiDARs are suddenly in the streets, reasons explained)

  • How to use a "wavelength trick" taught to me by mobileye's Engineers to help LiDARs see through rain and fog (ever wondered why some LiDARs aren't affected by weather conditions while most are? There is a secret sauce, and it's explained inside)

  • How to read any LiDAR Datasheet "on-demand" (I'll go through 5 LiDAR datasheets and help you understand some core elements such as the azimuth angle, range, field of view, photon sensitivity, angular resolution, channels, and more...)

  • An In-Depth Exploration of solid-State LiDAR technology, including MEMS (microelectronical systems), OPAs (optical phased arrays), and how does that compare to Flash LiDARs

  • The Principle behind Time Of Flight measurement, and how a LiDAR measures an object's exact 3D Position in the space (we'll also see Coherent - Continuous Wave technologies)

  • Why Pulse Time-Of-Flight LiDARs are blinded by fog, and why using a Continuous Wave system like AMCW LiDARs works better (hint: could continuous waves have more chances to go through than pulse waved?)
Still on fog:
  • The secret of the Dual Return technique, taught to me by Ouster, that helps LiDARs see through FOG and through obstacles
     
  • Why Dual Return doesn't work alone to see through fog, and the ONE true secret sauce companies use but don't advertise

  • How to know if a LiDAR company is hyping BS technology, and the main elements to look out for in a LiDAR

  • The never-told story of how I sold a LiDAR expertise phone call for 500$ — and the one skill that could 5x your market's worth

Let's take a break for a second and comment this last point.

I am not a LiDAR freelancer, and I don't usually do paid LiDAR missions to companies — but once in a while, I'm approached by companies and investors looking to do some "paid due diligence" regarding technology or companies.

Imagine, for example, if an investor came to you, and asked you "Can you tell me if this company is worth investing 1M$ in?".  This situation happens more often than you can imagine, and I was actually approached with this exact demand a few years ago. 

In this Module 1 of the course, I'm going to share the fully anonymized story of how I got hired for 500$ for a single phone call, and exactly what I had to know, and the pitfalls I had to overcome to get the client happy. This story has never been shared anywhere else outside of this course.

Finally, module 1 teaches way more things, such as:

  • What an FMCW LiDAR file looks like — and how engineers encode any property such as point velocity reflectance, or intensity, in a 3D Visualizer.

  • How to Visualize and Process a Point Cloud, and what you need to do to get a realistic visualization process

  • An analysis of LeddarTech's Flash LiDAR technology, and why Flash LiDARs could replace most other AV LiDARs in the future

  • The secrets behind FMCW LiDAR technology acquired after using products from companies like Aeva, bitsensing, mobileye, Blackmore, and others

  • Mobileye's FMCW LiDAR Technology explained, and the only possible way they can achieve true 4D.

  • ⚡️ LIDAR ENCODING PROJECT: In this first project, you'll analyze a Point Cloud file, and learn to encode information such as intensity, reflectance, or velocity of the point clouds to then visualize them.

    Here's an example with reflectance (notice how some elements are more red than others):

  • And a whole lot more, including How multi-channels LiDARs work, and the difference between a 2D LiDAR, a 3D LiDAR, and a 4D LiDAR; A simple formula you can use to calculate the number of points per second a LiDAR can emit, and how many points are needed for a self-driving car scenario; The LiDAR MindMap overview, and more...
Alright! I told you there would be a lot of new things, but this industry knowledge is not only incredibly valuable... it's also VERY HARD TO FIND. I had to gather dozens of research papers, datasheets, blog posts, technical reviews, paid surveys, and of course learn from all of these startups for years just to build this module 1.

After you've build the LiDAR skills, you will get the opportunity to build the LiDAR skills, so let's take a look at module 2:

MODULE 2

Point Clouds Processing & 3D Machine Learning

In the second module, we'll dive into Point Clouds processing. We'll learn how to visualize, encode, process, and build point clouds processing projects such as 3D Lane Line Detection or Driveable Area Segmentation.

What you'll learn:

  • The 10 Most Important 3D Machine Learning Algorithms every LiDAR Engineer must know, (and the most important of all in autonomous driving)

  • The only 3 ways to connect to a LiDAR's data after you've plugged it, and how to setup a LiDAR in a prototype self-driving car

  • The Top 5 Point Cloud Visualizers in the market, and how the Think Autonomous Team built its own 4D Perception Visualizer (plus, we'll also cover how to do Visualization on Jupyter based environments — which has been a real pain to LiDAR Engineers)

  • Why Open3D is NOT replacing the Point Cloud Library in most autonomous driving startups, and the 3 most important elements a self-driving car LiDAR library must have

  • 3 Libraries for 3D Deep Learning, and which one I recommend the most

  • How to use a LiDAR with ROS (including 2 demos of Self-Driving Car ROS Recordings used with LiDARs)

  • How to implement ghost, features, and region filtering on Point Clouds (including how to build a "bubble of security" to break a self-driving car if any object — even unclassified — ever comes within a radius of a car)

  • ⚡️ 3D LANE LINE DETECTION PROJECT: Use LiDAR's Reflectance Properties to identify traffic signs and lane lines, and build a 3D lane line detector

Lane Line Detection on LiDARs? Really?
Did you ever drive on a highway at night? I remember recently driving on the highway for over 3 hours in the plain obscurity at 120 km/h speed. What kept me going? What keeps anybody driving at night without light? It was the lane lines reflecting.

Cameras don't see during the night, but LiDARs do, and they even see lane lines. In this first project, we're going to implement a secret technique a former colleague taught me about how to leverage LiDARs to detect and fit 3D Lane Lines.


We'll then continue:
  • 5 Unsupervised Learning techniques you must know to build any 3D Machine Learning project

  • How to Voxelize a Point Cloud environment, and how this feature can be used for 3D Deep Learning, or Downsampling.

  • How to do 3D Segmentation using the RANSAC Algorithm (disclaimer: although advertised as such, RANSAC isn't a segmentation algorithm, but an outlier detection algorithm, and it therefore can be used in many more ways, we'll see a few of them too)

  • An in-depth look at the most used 3D Clustering approaches for Point Clouds (we'll see how to use these techniques for 3D object detection, 3D segmentation, lane line fitting, and many more)

  • The secret KD-Tree technique to 10x the speed of your LiDARs algorithms

  • How to use unsupervised learning to fit 3D Bounding Boxes to objects, and how to "predetermine" a bounding box based on your Point Clouds shape
This second project of the module was the main project of the v1 of this course, but it has been refreshed and improved for the v2. At the end of it, you'll be able to detect objects in 3D, find 3D drivable surfaces, and do many more...

And if this wasn't enough: There is a third, additional module, I decided to include based on some requests from LiDAR companies I met, who told me my course should have that.

This module is much less known, much more "field-level" as well. By this, I mean that this entire module has been built based on what companies told me they were working on, looking to learn, or implementing in their self-driving cars.

In a way, this is the kind of knowledge that can help you break the invisible barrier between experts and learners, and finally help you be identified as "one of them".

MODULE 3

Advanced Point Clouds Processing Concepts

In this third module, we'll see feature extraction, point cloud registration, and surface reconstruction on Point Clouds.

What you'll learn:

  • 12 Types of Point Cloud Features you can use to collect relevant information, even if your LiDAR is only a 2D XYZ LiDAR

  • How to compute surface normals, and a look at the most used algorithm in point cloud's research (including a C++ analysis of a surface normals estimation code)

  • 3 Most Important families of Keypoint Estimation algorithms, and an analysis of the Point Cloud Library's algorithms implemented for Keypoint estimation, including Computer Vision algorithms adapted to Point Clouds

  • Solid-State Fusion: How to "fuse" several LiDARs together, and do "point cloud stitching", to get a 360° view from multiple LiDARs

  • How the FPFH Feature Description algorithm works, and what to do with manually computed features

  • 4 Point Cloud Registration Techniques, and how to implement the Iterative Closest Point in SLAM.

  • How to use Keypoints for Pose Estimation, and 3 things that make pose estimation on Point Clouds better than on cameras

  • How Surface Reconstruction works, and 3 secret algorithms used to build 3D objects out of points.

  • 5 Mini-Projects you can build using these concepts, including an advanced feature extraction technique to improve 3D Lane Line Detection

  • And many, many more...

DLC

POINT CLOUDS CONQUEROR: Ground Truth Ops

As an optional exploration, you can also get this course with a "DLC" — an advanced extra course teaching you the secrets of Professional LiDAR Processing pipelines.

  • Solid-State Fusion & LiDAR Stitching: How to Fuse Several LiDARs together no matter their types, dimensions, or number of points

  • Deep Dive into 5 Point Clouds Registration Approaches, and a cheatsheet to know when to use which (hint: regular ICP doesn't work)

  • Inside a Professional LiDAR Pipeline, and the #1 most used LiDAR Alignment technique

  • Implement a Multi-LiDAR Calibrator on real self-driving car recording (up to 3 LiDARs)

  • And many many more, including LiDAR clock synchronization, ghost filtering, noise removal, KISS-ICP preprocessing, and all the professional tactics you'll need as a LiDAR Engineer!

Get more Details

FAQs

What are the prerequisites to follow the course? And what is the format like?

This course is one of my most accessible technically. I would even say it's open to beginners. Maybe not all beginners can take it, so be sure you can at least code in Python, and have an idea of the most basic Machine Learning concepts, such as classification, regression, clustering, or the related algorithms.

This course is made longer than its previous v1, because it contains a lot more content and field concepts. Therefore, you should expect 5-7 hours to complete it.

You don't need a LiDAR to do the course, and you can use any computer. The format is made of text, videos, projects, code, secret tapes, and PDFs, and most of the course will be done locally or on Google Colab...

Are there jobs in the LiDAR space?

Let me tell you something. Back in January 2023, I travelled to Las Vegas to attend the Consumer Electronics Show (CES), where I had several meetings planned (including with LiDAR startups). So on day 1, I went to the "West Hall"  where all the autonomous tech was, and guess what I found?

LiDARs, LiDARs everywhere...

They could have renamed the West Hall the "LiDAR Pavillon" and nobody would have complained. From LiDARs used by Netflix, to LiDARs used by the self-driving car industry, the LiDAR market is BOOMING.

Yet, the number of engineers is growing linearly. A few engineers coming out of school and deciding to specialize in LiDARs. Just a handful of them, year after year. This creates a serious GAP between the increasing demand for LiDAR skills, and the actual supply of LiDAR engineers.

LiDAR profiles are rare, in-demand, and valued in the market space, but it doesn't mean anyone who takes this course will automatically become a LiDAR expert. Neither does it mean anyone taking this course will automatically be paid 100k on their first job. Just like everything, it will demand you WORK, PATIENCE, and WILLINGNESS to overcome FRUSTRATION. 

But frustration is nothing else than your brain learning, and I believe those who will spend the time on these skills will have much better chances of success in the LiDAR industry.

How much are LiDAR skills worth?

How much is your current knowledge worth? The reality is, you may already possess valuable knowledge and not even know about it. And you may also not possess some "low hanging" skills that would be valuable to you, but not being aware of it.

Earlier in this page, I shared how I got paid several hundreds for a LiDAR phone call. But I wasn't even aware my skills were worth this much when I did the contract — I got this number suggested by the clients, which implies it could have been much bigger.

The goal of this course to bring you the "field knowledge" — the type of knowledge that makes you know what is valuable, what isn't, and that helps you communicate with recruiters or industry's experts that you are "one of them" — and not just an outsider.

I already know some LiDAR concepts / have been through the v1, will I still learn things?

As for any course you ever take, there might be concepts you already know. Yet, you may not have seen these concepts in the context I'm going to show them to you. You may already know about an algorithm, without suspecting it can be used at a larger scope.

It's up to you to determine whether you have to learn the remaining skills on this page, but I wouldn't base my judgment just of how much I already know (do the opposite).

☄️

How is this course unique?

This course is unique because it's connected to the LiDAR market. It's been the result of over 4 years of work on LiDARS, accumulating the "field knowledge" you can only get by working in the field.

When looking for tutorials and other courses, you might see a traditional approach that will show you how to work with 3D Data, or how to do "3D meshes", or 3D this, 3D that, but without having that connection to the field.

Yet, being connected, because it helps you identify the skills that are in-demand, the projects currently ongoing, the claims companies are fighting over, and therefore be an active member of the LiDAR industry, rather than just skilled with 3D data.

To picture it differently:

📼 Imagine you have 30+ LiDAR companies "bugged", and that knowledge processed and synthesised to help you understand how LiDARs really work, and what startups in the field really need.

But how much do companies really share publicly? Companies share a lot, and they share nothing. A single talk with a single company might not get you much, but it's when combining all the talks together, and when combining this knowledge with technical expertise (research papers, optics knowledge, physics, ...) that you can sometimes "get" what companies really mean. 

Do this at an intense level, and you have some elements of this course.

Testimonials

Mayur Waghchoure, Autonomous Driving Software Engineer, Dorle Controls

Recently, I have completed a course on introduction to 3D Perception from Think Autonomous; I am extremely satisfied with the content and your delivery. I have started with a new course on 3D Deep Learning and now learning PyTorch.

Sujitha Kurup, Research Assistant at SVKM's Narsee Monjee Institute of Management Studies

I am a research scholar based in India, trying to work on 3D modeling of the indoor environment. I was running in circles regarding my point cloud until I came across your website to join the course on 3D Perception and since then it has been a really great learning experience.

Jay Reddi - Senior Specialist, Deloitte

I really enjoyed the course! Thank you for preparing clear and concise lectures and interesting coding assignments. I took this course to get a better understanding of lidar in general and learning about Open3d and pptk was very useful. 

Varsha Sathya, System Engineer - Autonomous and Connected Vehicles, Tata Consultancy Services

I was absolutely new to point cloud and was quite lost in the information available online. This course is such a good kick-starter for anyone who wants to start upskilling with 3D perception. Also quick support from Jeremy and the team from simple doubts to any other issues is commendable. This makes the course more unique as its not just a bunch of videos but also mentor level support from industry experts.

Reshma K, PhD in 3D Object detection for autonomous vehicles

"The price was originally a blocking point for me, but it was really informative from a practical point of view."

"As a PhD student who works on data fusion and has not worked much on lidars before, the practical perspective given to point cloud processing in this course would definitely help get a head start on understanding and coding with point clouds.

I like that it is very hands on and the results are able to be visualized at every step. So, there is an instant gratification there, which gives you the motivation to go forward. Majorly, you get an insightful understanding. I like the part where you explain the functioning of inbuilt functions because in my field of work, it is important to understand the inside out and not just the end result.

I would definitely recommend this product to beginners like me because if you are working on lidar point cloud processing in a deadline based environment, be it industry or academia, and if you have never worked on lidars before, you definitely need a head start with the structure and understanding of point clouds, else you will be lost at sea if you delve without an anchor point. This course provides you the skeleton and you can build on from there."

Rahul Guptha, Point Clouds Conqueror

I'm halfway through the "POINT CLOUD CONQUEROR" course and I must say each module has been full of surprises and interesting.

"It's been a fantastic experience so far. I have chosen this course because it aligns perfectly with my career goals in Robotics. What I loved most about it was the Mind map summary in each module and hands-on approach.

Overall, I would highly recommend this course to students who are looking to gain deep insights into LiDAR and processing techniques.

Thank you once again for providing me with the opportunity to learn new things."

PICK YOUR LIDAR ADVENTURE

Which Point Clouds Conqueror do you need?

299€ or 2 x 149€

CONQUEROR

  • POINT CLOUDS CONQUEROR: 299€

—

299€

Buy Now

399€ or 2 x 199€

OPS EDITION

  • POINT CLOUDS CONQUEROR: 299€

  • DLC: GROUND TRUTH OPS: 199€

    —

499€ 399€

Buy Now