<< Return to Homepage

Interesting Tech Videos

The aim of this page is to gather some interesting videos for me to watch on topics of autonomous vehicles, machine learning, robotics and entrepreneurship, while hoping they have a general interest to the public. If you have any comment, please feel free to raise an issues on Github here, or star/watch my repo if you like it and want to keep an update. Please note, the shared videos and my comments represent my personal views.

Last update on Nov 26, 2023


1. AI and Software
2. Cars
3. Robotics
4. Automotive Industry

1. AI and Software

Title Author Affiliation Videos Comments
Intro to Large Language Models Andrej Kapathy OpenAI Good tutorial
Opportunities in AI 2023 Andrew Ng Stanford Generative AI is changing the entire landscape of computing.
Event-Driven Architecture Mark Richards "Developer to Architect" The entire lecture series from this website is very inspiring. The author also wrote a book called "Fundamentals of Software Architecture".
Data-centric AI: Real World Approaches Andrew Ng Landing AI Code/Model is fixed but improve the data.
Nvidia's massive GTC 2021 press conference in 17 minutes Jensen Huang Nvidia Computing is pushing all the boundaries
AI Infusion & Investment Opportunities at AAAI 2021 Kai-Fu Lee Sinovation https://slideslive.com/38952432/ai-infusion-investment-opportunities Curious to see the US apps copied from China and more insights about the AI transform and large data mining from the market data.
Laplace’s Gremlin: Uncertainty and Artificial Intelligence Neil Lawrence Cambridge It's a keynote so everything is presented in a rather abstract level, but might be worth listening like a TED talk.
Self-Supervised Learning & World Models - ICRA 2020 Yann Lecun and Wolfram Burgard FAIR Interesting discussion about robust DL
Particle-based Inference for Bayesian deep learning Jun Zhu Tsinghua Prof. Zhu's work has been circulating around probabilistic ML to develop robust ML and understand DNNs. His team has been working with Bosch AI on probabilistic ML. His page can be found here.
Understanding deep networks and its role played in prioritized search Yuandong Tian Facebook Research Interesting talk and wrap-up on understanding DNNs.
Training Generative Adversarial Networks with Limited Data Tero Karras NVidia It's a fantastic work done in NVidia Helsinki which can generate samples from small data. The application of face morphing (transition between facing) has been showcased in "Oodi" the major library in Helsinki.
Noise2Noise: Learning Image Restoration without Clean Data Jaakko Lehtinen NVidia, Aalto It's one of the most interesting ML seminar talks organized by FCAI.
Binarized Neural Networks on microcontrollers Lukas Geiger Plumerai They have an interesting NeurIPS paper on binary nets recently.
OpenAI GPT-3 Two-Minute Papers

I simply like this video channel of Two-Minute Papers. It's a nice way to grasp the progress of the latest work.

The Real AI Revolution Chris Bishop The author of textbook "Pattern recognition and Machine Learning", Microsoft https://nips.cc/virtual/2020/public/invited_16165.html

Keynote at Neurips 2020. Fundamental and applied research are getting mixed.

Component wise approximate Bayesian computation via Gibbs like steps Christian Robert Invited talk at Criteo AI Lab The series of invited talks by Criteo has a strong Bayesian taste, which I suppose is helpful for ad retargetting.
The Promised Land of AI Dr. Demis Hassabis and Prof. Amnon Shashua. Dan David Prize Event Great to hear about the research career of Prof. Shashua. More talks from him can be found here.
Domain-Specific Hardware Accelerators Bill Dally NVidia

The computing future lies in the edge with a surge of IoT devices. How do we develop algorithms following the trend of domain-specific chips?

Ultra Efficient AI on Resource Constrained Compute Platforms Mohammad Rastegari Formerly XNor AI, now acquired by Apple It's a good work on deploying intelligent algorithms on low-end devices. There remains many potentials for improvement.
Google Brain AutoML Quoc Le Google Quoc Le's work has been revolving around making learning automatic by feeding in a large amount of data, which was originated from his phd.


<< Return to Homepage

2. Cars and Autonomous Driving

Title Author Affiliation Videos Comments
Smart Eye Automotive Interior Sensing Smart Eye Smart Eye is a pioneer in Driver Monitoring Systems (DMS). With a requirement from EURO NCAP and its extension to occupant monitoring, we see a growing number of players into the IMS (interior monitoring system) market, including Tier-1's, AD companies (MobilEye and Zenseact), chip companies (Qualcomm), and conventional eye-tracking companies (Seeing Machines).
A Path Towards Autonomous AI Yann LeCun Meta ML research moves to System 2 in reasoning and cognition. Slides. Blog.
Mobileye's Lean Driving Policy Shai Shalev-Shwartz MobilEye Good to see an integration of Responsibility Sensitivity Safety (RSS) and Markov Decision Processes (MDP).
Under the hood 2022 Amnon Shashua MobilEye Once a year now at CES 2022, we will see the progress made by MobilEye.
Challenges of machine learning for autonomous driving Patrick Perez Valeo.AI Good to see the progress by the Lidar company
Tesla AI Day Highlights 2021 Lex Fridman What I would like to add is that Tesla's engineering solution is problem-driven, while other car OEMs are more of keeping up with the peers. No matter if they succeed or not eventually, the intellectural assets (methods and techniques) they develop is valuable for the future of AI development, which will be included in the textbooks.
CVPR'21 Workshop on Autonomous Vehicles Andrej Karpathy Tesla Andrej almost gives a presentation at CVPR every year, and good to see what is the update, and here follows my comments.
  • Tesla makes a distinctive approach towards self-driving in vision only, in-house neural network design by guiding its SoC design and supercomputer setup, and massive data collection. It is an overall good approach concern is whether the validation and verification of their function is sufficient.
  • The replacement of radar or any other types of sensors is the pain point when it comes to sensor fusion. Usually a confidence-based late fusion approach is adopted but maybe a more complex hybrid approach with vision to give a final justification can be used. The tradeoff is the computational cost but the car will be equipped with a complementary set of sensor set.
  • Vision only (without Lidar to tackle the harsh weather) means they will have a narrower application domain or you just have to be brave.
  • I agree that vision only approach is more scalable in terms of generic decision making but the car will always encounter unknowns and "surprises". Not to mention, it will save a huge amount of cost in maintaining an HD map.
  • Automatic data labeling is smart with more work done offline and massive data collection in the shadow mode.
  • 221 triggers is a good starting point for fucntion design and verification.
  • Data engine has been talked about for a long time and VW seems made their "big loop" version of it.
  • Compute is the new horse power for cars. Innovative chips scaling from supercomputer to in-car processors will be the very core for Tesla which is tailored by the design of ad-hoc neural nets.
  • With the contraints of data regulations in China, I wonder how this approach would work and prevail. It requires daily communications and updates in their function design (neural network training) to tackle the complex Chinese traffic. Also, the local competitors have a wider access to software/AI competence in designing even superior neural networks. SoC is a big hype there too and supercomputer is a national interest.
  • David Silver made some interesting comments here.
CVPR'21 Tutorial: All About Self-Driving Raquel Urtasun Waabi Prof. Urtasun started her company after Uber with her former students, and this series of talks is mainly the outcome from their experience including topic on perception, prediction, motion planning, hardward, v2v, simulation, and hd maps. More info here.
HD Live Maps for Automated Driving: An AI Approach Xin Chen HERE This was a talk held in 2018 at Chicago Uni, and it's interesting to see HERE's work on HD map generation.
CVPR'20 Workshop on Scalability in Autonomous Driving Keynote Andrej Karpathy Tesla He presented a neural network solution for autonomous driving in an academic setting. I also liked the comments in this article which talked about unreasonable effective of data.
Testing Apollo Autonomous Vehicles at Night and Other Tricky Scenarios Baidu Apollo There are two things to check out in this video. One is how complex the China scenarios can be, and the other is how to adept to them by the Baidu Apollo System.
What Do AVs Mean For Infrastructure? Partners for Automated Vehicle Education It's quite interesting to learn how Intelligent Transportation System and road infrastructure would affect AVs and their reversed impacts. The entire channel is quite interesting.
Probabilistic Robotics and Autonomous Driving Wolfram Burgard University of Freiburg Interesting panel hosted by Toyota Research Institute.
System Design for Autonomous Vehicles with Drago Anguelov Drago Anguelov Head of Research at Waymo Many insights from a perception/ML expert. The audio podcast can be found from here.
A future with affordable Self-driving vehicles Raquel Urtasun Professor at University of Toronto and head of Uber ATG

We hardly find an online video by Prof. Urtasun before the pandemic due to the IP with Uber. Now after a clear takeover of Uber self-driving units by Aurora, many curious unknowns are released, though we could read about their work previously from her website. It's a shame that she did not get an offer to move to Aurora as their future development will be less transparent. A video channel by the professor about many topics on self-driving cars can be found here.

Mathematical Challenges and Opportunities for Autonomous Vehicles 2020 Institute for Pure & Applied Mathematics (IPAM) UCLA This is a wonderful list of AI talks in the context of self-driving cars. My personal take is driving is a very complex task. It cannot be simply decomposed into perception, planning and control, as we human learn to run and navigate from ancient times and they are hardwired in our brain. Replacing human drivers is impossible without a mature algorithm under investigation by DeepMind.
ECCV 2020 Workshop on Map-based Localization for autonomous driving Technical University of Munich The entire session of the workshop covers the real-time localization by maps, and both map generation and updates are discussed.
AI for Full-Self Driving at Tesla Andrej Karpathy Tesla Karpathy is the main drive of the perception stack at Tesla. They boast the end-to-end driving with cameras only, in line with the early development by MobilEye.
Waymo and the Future of Self-Driving Cars Dmitri Dolgov CTO of Waymo

It's nice to hear about the evolution of Waymo from a participant in DARPA challenge to a commercial company (profittable some day perhaps). I still think with all the sensors, simulation, testing and other efforts, the functional boundary ("ODD") for L4 will be very limited. But it's interesting to see Waymo as a testbed of DeepMind's new algorithms.

Automated Driving Systems – Measurable Safety: Standards, Regulation and Pragmatic Solutions Gil Amid Foretelix

An Isreali firm who collaborates with the German OEMs for the rollout of autonomous vehicles. I strongly believe vehicle satefy should not be taken as a competing edge by car manufacturers. The development in this area should be shared and transparent.

Baidu Apollo Test Ride

A recent news says Baidu is going to make their own intelligent EV together with Geely holdings. The implications are two-fold in my opinion. The tech company starts emerging into the auto industry, as the ecosystem changes from a traditional hierarchical supply chain with OEMs on top, to a triangular shape between tech companies, tier-1 suppliers and car OEMs. With their strength in IT (digital services) and AI (image recognition, natural language processing), there is a list of areas they could prevail in making cars, such as automated driving function, intelligent cockpit, and 5G connectivity.

But will tech companies succeed in making their own cars? It's still too early to say. Cars are complex systems even with reduced mechanical complexity thanks to the increase of electric components, and passenger safety is critical. Similar to making phones by tech companies in recent years, there is a set of hurdles they need to cross. What I am interested to see is its similarity to Waymo of Google. The latest advances in machine learning from DeepMind and numerous products from Google, ranging from Assistant, search engine and visual surveillance, will be deeply integrated to their cars.


<< Return to Homepage

3. Robotics (Perception, control and learning)

Title Author Affiliation Videos Comments
Reinforcement Learning: Past, Present, and Future Perspectives Katja Hofmann Microsoft Research Nice tutorial on reinforcement learning at Neurips 2018
From SLAM to Spatial AI Andrew Davison Dyson and Imperial College

The talk was given by the one of the original authors of SLAM. The entire video channel of Robotics Today is quite interesting.

Doing RL on Robots Jan Peters TU Darmstadt There are two people to watch in Robot Learning in my opinion. One is him and the other one is Sergey Levine. His lab and former students have a large impact on the auto industry and manufacturing in Bavaria area of Germany.
Robots Learning (Through) Interactions Jens Kober Bosch AI Lab BCAI has a branch in Amsterdam and the machine learning lab is the other one to watch.
Introducing the Zenmuse L1 DJI

It's an amazing product by combining Lidar seamlessly with a drone.

Opportunities and Challenges for Autonomy in Micro Aerial Vehicles Vijay Kumar UPenn Prof. Kumar has been a pioneer in UAV's for the last decade. The drones developed from their technique have many use cases and potentials.


<< Return to Homepage

4. Automotive Industry and Beyond

Title Author Affiliation Videos Comments
Interview with Elon Musk TED
Interview with Peter Mertens Peter Mertens Former head of R&D at Audi, VW, Volvo, Jaguar, etc

It is a second interview with the car veterain. Since he got sick and didn't returned to car OEMs, he dares to point out the pain points in the traditional OEMs - short of budgets but a lot to invest in electrification and software, no resource, detrimental old mindsets in horizontal integration, etc. His first interview is even thornier which sometimes I think it should not be spelled out. The shared video is overall a very insightful interview if people want to know the car industry as of now.

It will be a painful journey for traditional OEMs like VW and Mercedes-Benz to catch up in the development of OS, and rapid expansion of EV's. For example, there is a big communication gap between the management and developers, and traditional car managers simply are not capable of taking a bottom-up approach. While Tesla boss can be called on a Sunday to resolve problems, traditional managers don't understand technical details and developers don't get the big picture. Consequently, the strategy made by the managers can be misleading while others don't have a clue. We will see a large number of consolidations in OEMs (like PSA and FCA) in the next few years to weather the storm of vast spendings and market shrinkage in the area of their conventional strength.


<< Return to Homepage

© 2021-2022 Guanqun Cao. All Rights Reserved.