Defining and measuring decision-making styles, also known as playstyles, is crucial in gaming, where these styles reflect a broad spectrum of individuality and diversity. However, finding a universally applicable measure for these styles poses a challenge. Building on Playstyle Distance, the first unsupervised metric to measure playstyle similarity based on game screens and raw actions by identifying comparable states with discrete representations for computing policy distance, we introduce three enhancements to increase accuracy: multiscale analysis with varied state granularity, a perceptual kernel rooted in psychology, and the utilization of the intersection-over-union method for efficient evaluation. These innovations not only advance measurement precision but also offer insights into human cognition of similarity. Across two racing games and seven Atari games, our techniques significantly improve the precision of zero-shot playstyle classification, achieving an accuracy exceeding 90% with fewer than 512 observation-action pairs--less than half an episode of these games. Furthermore, our experiments with 2048 and Go demonstrate the potential of discrete playstyle measures in puzzle and board games. We also develop an algorithm for assessing decision-making diversity using these measures. Our findings improve the measurement of end-to-end game analysis and the evolution of artificial intelligence for diverse playstyles.

Torc Robotics and Daimler Truck AG enter their fourth year of partnership, with a focus on customers, industry collaboration, and commercializing Level 4 autonomous trucks in the U.S. for long-haul applications. The powerhouse team continues to develop safe, sustained innovation in the freight industry as they combine Daimler Truck's extensive experience in manufacturing and relationships in the freight industry with Torc's experience in developing autonomous vehicle solutions. Since Daimler Truck's majority share investment in Torc in 2019, the two have worked hand-in-hand to be the first to commercialize a profitable autonomous truck solution at scale. Torc continues to operate as an independent subsidiary and serves as the lead for autonomous driving system development, innovation, and fleet testing. "Bringing a safe Level 4 autonomous truck to market is by no means a simple task. Over the past three years, we have benefited from the strong collaboration with Daimler Truck, bringing us significantly closer to our goal of developing a highly optimized self-driving truck that will meet the fleets' needs for cost, safety, and performance. The teamwork shown has been outstanding so far and we're entering our fourth year of partnership with a clear roadmap – focusing on one manufacturer and one initial use case in one geographic area," said Peter Vaughan Schmidt, Torc CEO.

At Torc, we have always believed that autonomous vehicle technology will transform how we travel, move freight, and do business. A leader in autonomous driving since 2007, Torc has spent over a decade commercializing our solutions with experienced partners. Now a part of the Daimler family, we are focused solely on developing software for automated trucks to transform how the world moves freight. Join us and catapult your career with the company that helped pioneer autonomous technology, and the first AV software company with the vision to partner directly with a truck manufacturer. This is a team of experienced engineers & Technicians that work on our autonomous trucks to integrate hardware and deploy software.

Torc Robotics and Daimler Truck kick off their third year of partnership poised to commercialize the first scalable, profitable Level 4 autonomous truck that will help fleets improve their operations while bolstering the backbone of the U.S. economy. Torc is currently testing the Level 4 trucks on public roads in Virginia, New Mexico, and Texas, with continued route expansion in the works. The two companies are pursuing a focused, safety-oriented approach to market that also seeks to build trust among fleets and the drivers of vehicles who will share the road. Introducing a world-changing technology into an existing infrastructure, where human drivers will share the road with automated trucks, requires credibility and responsibility, according to Dr. Peter Vaughan Schmidt, Head of Daimler Truck's Autonomous Technology Group. "As the inventor of the truck, Daimler Truck has many decades of experience in testing and validation of commercial vehicles. Nevertheless, to develop a safe autonomous level 4 truck remains a complex task and resembles a marathon, not a sprint. Two years together with Torc Robotics, we have accomplished a lot, collaboratively pursuing a common goal of leading the logistics sector into the future and making road traffic safer for society. I am convinced that we are optimally positioned as a company and together with Torc we have the right partner at our side to achieve our goals."

Daimler is clearly eager to expand its plans for self-driving trucks. The automotive giant is teaming up with Waymo to develop trucks capable of level 4 autonomy, or full self-driving in specific conditions. The early strategy will focus on a modified Freightliner Cascadia that uses Waymo Driver for navigation. This first truck will be available in the US in the "coming years," the companies said. The two would also "investigate" expanding their efforts to other brands and markets.

Deep reinforcement learning (DRL) has achieved great success in various applications. However, recent studies show that machine learning models are vulnerable to adversarial attacks. DRL models have been attacked by adding perturbations to observations. While such observation based attack is only one aspect of potential attacks on DRL, other forms of attacks which are more practical require further analysis, such as manipulating environment dynamics. Therefore, we propose to understand the vulnerabilities of DRL from various perspectives and provide a thorough taxonomy of potential attacks. We conduct the first set of experiments on the unexplored parts within the taxonomy. In addition to current observation based attacks against DRL, we propose the first targeted attacks based on action space and environment dynamics. We also introduce the online sequential attacks based on temporal consistency information among frames. To better estimate gradient in black-box setting, we propose a sampling strategy and theoretically prove its efficiency and estimation error bound. We conduct extensive experiments to compare the effectiveness of different attacks with several baselines in various environments, including game playing, robotics control, and autonomous driving.

Autonomous-driving company Torc Robotics may not be as well known as, say, Waymo, but that may change soon as Torc looks to expand. The company is looking to nearly double its number of employees in order to continue developing tech for self-driving cars. Torc unveiled its Asimov (named after science-fiction writer Isaac Asimov) autonomous-driving system last year, and gave public demonstrations at CES 2018. The company is headquartered in Blacksburg, Virginia, and continues to test self-driving cars there and in Las Vegas. Last year, it sent one of its modified Lexus RX SUVs on a cross-country trip. But the company's history goes back further.

For the past decade, the easiest way to spot a self-driving car was to look for the distinctive spinning bucket mounted to its roof. The classic lidar design pioneered by Velodyne spins 64 lasers through 360 degrees, producing a three-dimensional view of the car's surroundings from the reflected laser beams. That complicated and bulky set-up has traditionally also been expensive. Velodyne's US $75,000 lidar famously cost several times the sticker price of the Toyota Priuses that formed the nucleus of Google's original self-driving car fleet. Those days are long gone.

Self-driving cars are getting closer and closer to become an everyday reality. Although, at first it may seem like that autonomous cars investigations are reserved for a very narrow group of researchers, we would like to show it is not necessary true. Actually, the only things you need to start playing with driverless-cars, are some hacking skills, a little bit of programming and basic understanding of machine learning concepts - mainly deep and reinforcement learning. Driverless cars have been a dream of engineers since automotive industry was born and the first approaches were made, when Ford Model T was still ruling the roads. Although, radio-controlled car, presented by Houdina Radio Control in 1925, is far away from what we understand as an autonomous car in 21th century, it might be considered as the first try to construct an automobile, that does not require a human behind the wheel.

In the previous blog post Using Keras and Deep Q-Network to Play FlappyBird we demonstrate using Deep Q-Network to play FlappyBird. However, a big limitation of Deep Q-Network is that the outputs/actions are discrete while the action like steering are continuous in car racing. An obvious approach to adapt DQN to continuous domains is to simply discretize the action space. However, we encounter the "curse of dimensionality" problem. For example, if you discretize the steering wheel from -90 to 90 degrees in 5 degrees each and acceleration from 0km to 300km in 5km each, your output combinations will be 36 steering states times 60 velocity states which equals to 2160 possible combinations. The situation will be worse when you want to build robots to perform something very specialized, such as brain surgery that requires fine control of actions and naive discretization will not able to achieve the required precision to do the operations.