Amazon this week declared AWS DeepRacer, a completely autonomous 1/18th-scale race vehicle that plans to assist developers learn ML. having a price tag of $399 but presently provided for $249, the race car allows developers get access to a ML method dubbed as RL (reinforcement learning). RL takes a different viewpoint to train models as compared to other ML methods, Amazon claimed to the media in an interview.
It is a kind of ML that operates when an “agent” is permitted to respond within an interactive environment on a trial-and-error basis. It does so by employing feedback from those reactions to learn eventually in order to reach a prearranged objective or to increase some kinds of reward or score. This makes it stand out from other ML methods—such as SL (Supervised Learning), for instance— as it does not need any labeled training info to begin, and it can make short-term moves while working for a long-term objective.
On a related note, driverless car functionalities are getting closer to mass-market manufacturing with ARM (the British chip designer) launching the first safety-focused chips for developing features such as automated collision prevention into cars. The new line of “Automotive Enhanced”, or AE, application chipsets allows chipmakers develop processors with security functions that permit autonomous vehicles to meet the hardest safety needs, claimed the firm. The company announced this as it unveiled the Cortex-A76AE, its first autonomous chipset.
It anticipates the first cars employing the chip to hit the streets by 2020. Current ARM clients operating on autonomous driving platforms comprise NXP, Nvidia, Samsung’s Harman business, Renesas, and Siemens Mentor unit amongst others. Carmakers and their providers wish to be capable of running many different applications with different levels of safety needs all within the same processor, claimed vice president of ARM’s automotive business, Lakshmi Mandyam, to the media.