Lifeboat Foundation

At this year’s Conference on Machine Learning and Systems (MLSys), we and our colleagues presented a new auto-scheduler called DietCode, which handles dynamic-shape workloads much more efficiently than its predecessors. Where existing auto-encoders have to optimize each possible shape individually, DietCode constructs a shape-generic search space that enables it to optimize all possible shapes simultaneously.

We tested our approach on a natural-language-processing (NLP) task that could take inputs ranging in size from 1 to 128 tokens. When we use a random sampling of input sizes that reflects a plausible real-world distribution, we speed up the optimization process almost sixfold relative to the best prior auto-scheduler. That speedup increases to more than 94-fold when we consider all possible shapes.

Despite being much faster, DietCode also improves the performance of the resulting code, by up to 70% relative to prior auto-schedulers and up to 19% relative to hand-optimized code in existing tensor operation libraries. It thus promises to speed up our customers’ dynamic-shaped machine learning workloads.

Researchers at Oxford University’s Department of Computer Science, in collaboration with colleagues from Bogazici University, Turkey, have developed a novel artificial intelligence (AI) system to enable autonomous vehicles (AVs) achieve safer and more reliable navigation capability, especially under adverse weather conditions and GPS-denied driving scenarios. The results have been published today in Nature Machine Intelligence.

Yasin Almalioglu, who completed the research as part of his DPhil in the Department of Computer Science, said, “The difficulty for AVs to achieve precise positioning during challenging adverse weather is a major reason why these have been limited to relatively small-scale trials up to now. For instance, weather such as rain or snow may cause an AV to detect itself in the wrong lane before a turn, or to stop too late at an intersection because of imprecise positioning.”

To overcome this problem, Almalioglu and his colleagues developed a novel, self-supervised deep learning model for ego-motion estimation, a crucial component of an AV’s driving system that estimates the car’s moving position relative to objects observed from the car itself. The model brought together richly-detailed information from visual sensors (which can be disrupted by adverse conditions) with data from weather-immune sources (such as radar), so that the benefits of each can be used under different weather conditions.

It will be commercially available in early 2023.

London-based SolarBotanic Trees (SBT) unveiled its solar and storage system that is shaped liked a tree. The company aims to deploy its technology to charge electric vehicles (EVs) to begin with, Electrek.

With the world moving towards less-carbon emissions, there is a rush toward harnessing renewable sources of energy. Not only do these technologies need improvements in their power generation efficiencies, but they also need to be aesthetically pleasing.

Ray Kurzweil predicted Technological Singularity nearly 20 years ago. Elon Musk could enable a world of economic abundance with real world AI. Robotaxi and Teslabot will transform the world more than car and the first industrial revolution.

Tesla sells Model Ys for about $60000, but it currently costs them about $30000–40000 to make them. A Teslabot is 1/30th of the mass of a Model Y. It will use 1/30th of the batteries. The software is an overall cost of development. If billions of bots are produced then the cost would trend toward the cost of the hardware plus Apple iPhone-like margins including the software (say 40% gross margin). At Model Y cost of $30k then the hardware cost for Teslabot will go to $1000. $2000 with margins and software. A bot can work for 8,000 hours in a year. 8,760 hours in a year. $2000 divided by 8,000 hours is $0.25. If you add 10 cents per hour for electricity then it is $0.35 per hour. Going beyond that is bots can work in the factory and work cheaper than humans. Currently 15,000 workers in Tesla China factory. Replace all of them with $0.35 per hour bots. Reduce labor cost component. If a lot of bots can increase production rates. by 2X then all costs spread over more units. Bot-produced solar and batteries can lower the cost of energy by vastly increasing the supply. Those trends could get us to $500‑1000 per bot costs and lower energy costs. Having virtually unlimited labor costing less than 35 cents per hour will be transformational.

Continue reading “Elon’s Real World AI is the Real World Technological Singularity” »

A lot of hurdles remain before the emerging technology of self-driving personal and commercial vehicles is common, but transportation researchers at The University of Alabama developed a promising, inexpensive system to overcome one challenge: GPS hacking that can send a self-driving vehicle to the wrong destination.

Initial research shows a self-driving vehicle can use already installed sensors to detect traveling the wrong route when passengers are unaware of the change, thwarting an attempt to spoof the GPS signal to the vehicle, according to findings outlined in recently published papers in the IEEE Transactions on Intelligent Transportation Systems and Transportation Research Record: Journal of the Transportation Research Board.

Relying on software code and in-vehicle sensors already part of the self-driving system would be cheaper for consumer and commercial vehicles to deny the hacked directions used to steer cargo or people away from their intended destination, said Dr. Mizanur Rahman, assistant professor of civil, construction and environmental engineering and affiliate researcher with the Alabama Transportation Institute.

Security researchers have discovered that Minecraft is the most heavily abused game title by cybercriminals, who use it to lure unsuspecting players into installing malware.

Based on stats collected by the security firm between July 2021 and July 2022, Minecraft-related files accounted for roughly 25% of malicious files spreading via game brand abuse, followed by FIFA (11%), Roblox (9.5%), Far Cry (9.4%), and Call of Duty (9%).

Other game titles with notable percentages of abuse during this period are Need for Speed, Grand Theft Auto, Valorant, The Sims, and GS: GO.

The price has increased by $3,000, but Tesla’s FSD is still a work-in-progress.

Tesla’s Full Self-Driving Beta option now costs a hefty $15,000. Tesla CEO Elon Musk announced on Twitter late last month that it would increase the option’s price by $3,000.

Continue reading “Tesla’s Full Self-Driving now costs $15,000, but lawmakers argue it doesn’t live up to its name” »

Hydrogen (H 2) is currently discussed as an ideal energy carrier in a world requiring renewable energies. Hydrogen has the highest gravimetric energy density of all chemical fuels (141 MJ/kg), which is three times higher than gasoline (46 MJ/kg). However, its low volumetric density restricts its widespread use in transportation applications —as current storage options require a lot of space.

At ambient temperature, hydrogen is a gas, and one kilogram of hydrogen occupies a volume of 12,000 liters (12 cubic meters). In fuel-cell vehicles, hydrogen is stored under a very high pressure of 700 times the atmospheric pressure, which reduces the volume to 25 liters per kilogram of H 2.

Liquid hydrogen shows a higher density resulting in 14 liters per kilogram, but it requires extremely low temperatures since the boiling point of hydrogen is minus 253 °C.

AUSTIN (KXAN) — Walking around Austin, you may see something surprising — self-driving cars rolling around the roads.

Earlier this year, autonomous vehicle technology company Argo AI launched its driverless operation in Austin. Argo AI public policy and government relations manager Sly Majid said these cars are key to the future of transport.

“Autonomous vehicle technology is incredible,” he said. “The vehicle is doing the dynamic driving tasks; the vehicle is the driver of the car.”