Posted by Eric Huang on January 30, 2019
CES featured lots of automotive, AI and robotic innovation. To function in the real world the devices need sensors, and really vision sensors gathering all sorts of “visual” data. This can really include lots
ARC EV gives designers a head start
ARC EV (embedded vision) processors, when used with the safety enhancement package, are the only vision processors in the industry that are ASIL D Ready Certified. This means that automotive SoC designers are a step ahead of the competition when they start developing ADAS chips – they are further along the road to ISO 26262 certification than designers who are not using ASIL Ready IP.
One example of ARC EV customer success
Automotive IC vendor, Arbe Robotics selected DesignWare ARC EV processors to speed their chip to market. Arbe Robotics’ imaging radar is specifically designed for ADAS and autonomous vehicles. The high-resolution imaging radar can sense the environment at a wide 100-degree field of view in high-resolution at the highest reliability in various weather and environment conditions, including fog, heavy rain, pitch darkness, and air pollution. Their chip creates a detailed image of the road at a range of more than 300 meters (1,000 feet).It captures the size, location, and velocity data of objects surrounding the vehicle to enable safer driving and support level 4 and 5 autonomous operation. Arbe Robotics’ custom radar chip utilizes an extensive portfolio of Synopsys’ ASIL B and D Ready certified DesignWare IP, including the ARC EV6x processors and its software, MetaWare EV.
Synopsys wins comparison testing by customers for Power, Performance, and Area
Our customers consistently do head-to-head evaluations of the EV6x for power/performance/area and an efficient balance of the three. Unfortunately, comparing different vision processor IP is not simple. Bleeding edge IP solutions often haven’t reached silicon yet, and every implementation is different, making it difficult to calculate and compare power or performance between IP options. No benchmark standards exist for comparing CNN solutions. An FPGA prototyping platform might provide accurate benchmarks but not accurate power estimates.
Measuring power consumption is hard, Synopsys Zebu to the rescue
One way to calculate power consumption is to run a RTL or Netlist based simulation to capture the toggling of all the logic. This information, using the layout of the design, can provide a good power estimate. For smaller designs, the simulation can be completed in hours, but for large designs and CNN graphs requiring high frame rates, a simulation could take weeks to reach a steady state to measure power. Synopsys’ ZeBu Server to the rescue. It provides a tremendous benefit for analyzing and measuring power for both IP developers and SoC designers. The ZeBu server is the industry’s fastest emulation system for complete SoC designs. ZeBu can compute power accurately for hundreds of millions of clock cycles, e.g. Manhattan GPU frame, in hours instead of months. SoC designers can use ZeBu to tune power consumption of all elements in a system. There is a real risk in choosing IP that skips detailed power measurements in favor of estimating power through shortcuts using smaller simulation models, because any over-optimism becomes a power problem for you to figure out on your SoC.
For more info on the ASIL D Ready Certified ARC EV6x with Safety Enhancement Package: https://www.synopsys.com/dw/ipdir.php?ds=ev6x-safety-enhancement-package
For more info on Arbe’s adoption of EV6x with SEP: https://news.synopsys.com/2018-06-26-Arbe-Robotics-Selects-Synopsys-IP-to-Enable-its-High-Resolution-Imaging-Radar-to-Achieve-the-Highest-Automotive-Safety-Level-for-Autonomous-Vehicles
Here’s the original blog on the ARC ASIL D announcement:
“It’s no secret we’re facing a serious labor shortage in agriculture so seeing companies devote time and funding to new technology is welcomed,” Jim Bair, President and CEO of the U.S. Apple Association told the NewsHour via email. “
The labor shortage derives from both locals not enough choosing to work in agriculture and a guest worker program not suited to meet the needs of the U.S. agriculture economy.
Not having enough workers directly effects a farm’s ability to harvest specifically, and operate in general.
Specifically, for example, in Washington State during Zuchinni season a zucchini can grow up to a full inch in 1 day. If labor isn’t available, they can’t harvest the food at the time they need to. For apples, the workers need to walk up and down ladders with big buckets hanging around their necks to carry harvested apples
Robots can assist.
For harvesting, robot makers must address to challenges: perception and manipulation
For Perception:, the sensors on the fruit pickers need to be able to find the fruit/plant, and determine if it’s ready to be picked. (An ARC EV system could be integrated into chip)
For Manipulation, the fruit must be harvested and gathered without damaging and reducing the value of the fruit.
The team at Good Fruit Grower explains it best. They demonstrate it specifically with an Apple picker.
Based on that video from 2 years ago, it may be that some of your Washington or Michigan apples will be picked by a robot this year