Cleaning robots have huge commercial value in the era of epidemic

Phil Duffy, vice president of innovation at Brain Corp, talks to Nick Flaherty about the future of robots that can be used in public spaces. While the robot offers transformational value, it has so far been available only in tightly controlled operating environments.

Organized from – eenewseurope

Phil Duffy, vice president of innovation at Brain Corp, talks to Nick Flaherty about the future of robots that can be used in public spaces. While the robot offers transformational value, it has so far been available only in tightly controlled operating environments. San Diego-based Brain Corp has developed artificial intelligence software that system developers can use to build self-driving machines that can navigate safely and efficiently in public indoor spaces such as retail stores, airports, and hospitals.

Cleaning robots have huge commercial value in the era of epidemic

The demand for robotic cleaning systems has grown due to the Covid-19 pandemic. In April, Brain Corp raised $36 million for a total investment of more than $160 million to roll out its artificial intelligence technology globally, with robots trained by operators to follow one route at a time.

Phil Duffy, CEO of Brain Corp, said: “We started with neuromorphic computing research to study how the brain processes vision and how the brain learns, which gives us a technology that can be applied to robotics to solve the problem of navigation in complex environments.”

“We have over 14,000 robots, the largest scale operating in public settings, retail and airports. The retail and cleaning industries have adopted robots earlier than anyone, so they are somewhat prepared for Covid-19, and they are in the middle of the day 133% increase in usage. But there are places out of reach of robots, which forces staff to sanitize in person.”

The key is that data can show where the bot has gone, and customers can set their own compliance baselines to show how clean those places are.

“One of the reasons we got into retail is because of its complexity. Cleaning, inventory distribution and scanning — if we can solve retail, then we can get into that, and it’s going to be big. The robots we’re seeing today It’s self-driving robots that can navigate things like material handling or cleaning, and then move into robotic arms for industrial automation, which is a huge opportunity for robotics.”

How to do it

The technology uses humans to train robots, showing them which route to take so they can repeat. Phil Duffy: “The process of using the machine is integrated into the process of the store itself, so we make it easy for the operator to use it.”

“This requires machine learning, we’ve driven over 4 billion square meters, over 3.1 million hours and 4.2 million kilometers of self-driving time, we’ve seen edge cases from the network effects of large fleets of robots, the bigger the dataset, the more you The more edge cases are resolved.”

For example, infrared heaters used in supermarkets. When these switches are on, the heater flushes the sensor. The other is ghost pixels. The robotic platform is visible, and if it sees obvious obstructions, but isn’t connected to anything, it’s possible ghost pixels, reflections, especially when it’s always in one area. This usually means that the machine stops because it sees something and the reflection filter doesn’t work anymore.

There are also some very strange situations that require the human eye to solve the problem. In a store, a robot kept stopping for no apparent reason. The result is that the door to the store opens and litter is flying in the air, which the system sees as an obstacle. These edge cases, like reflections from different materials, objects sticking out of shelves, different shopping carts, etc., will help improve the AI ​​framework.

Some robotic systems are using UV-C light to do the cleaning, but the Brain Corp-backed system targets chemical disinfectants.

“The problem with UV-C is that it has a very slow dwell time, so even with high power, it stays there for a long time, and it doesn’t clean everything. Chemical disinfectants start from the moment it’s dropped It’s sterilized, so it’s suitable for mobile robots. You can spray as you move, and there’s no downtime.”

The AI ​​platform, called BrainOS, utilizes cameras and lidars to run on off-the-shelf hardware and multi-sensor arrays. The controller is a mix of Intel and Qualcomm’s ARM-based Snapdragon chips.

On the sensor side, the platform uses a single or dual lidar from a German supplier for larger systems that require a 10-meter range. Duffy: “We can automate a camera, but we do a full safety analysis, including size, weight, speed and environment. We use lidar for mapping, SLAM, navigation, and people detection. And then there’s a tilted laser Radar to create a ‘virtual bumper’ as well as ToF cameras. All these sensors are combined on map and navigation data that are trained to map the space and thus create the route.”

“We can pair the phone with the machine, go and choose the route—if the route is complete, there’s a problem, or it’s blocked, it sends a text message to the operator with a picture. It can be re-routed locally, it can leave and plan a return route. We’re working on the ability to update the map to prevent periodic blockages. We are now developing area fill – specifying certain boundaries, the robot calculates the fill area to figure out the best path to clean. “

“The brain computing module is the OEM’s equipment, including the Electronic wiring harness, our job is to integrate it into the machine, run the pilot line, calibration and end-of-line systems, and transfer it to the OEM for production and establish the production process and quality system.”

This is based on Ubuntu Linux and BrainOS security layers, followed by a full stack with navigation primitives, and comes with a set of features for the user interface (UI) and cloud connectivity over various networks, including 5G.

“We built our own hardware abstraction layer (HAL), we took a platform approach to firmware for the sensors and motor drivers, so we were hardware agnostic,” he said.

Duffy: “Latency is critical for security authentication, so it’s all handled by the motherboard. Even with 5G, some operations may be offline. The low-level deterministic logic of security awareness is black or white, the machine is either on or off. So, it doesn’t need to deal with high-level logic to stop. At a higher level, it has cross-check sensors, handles motion detection, etc., which allows us to avoid delays in processing.”

The framework is then integrated with smart buildings.

“The sensor system is built in and the robot has the ability to perform these functions, including: retail, inventory scanning, adding scanning towers, pricing, viewing lights-off operations, checking office occupancy levels, temperature control, WiFi and LTE coverage, and the usefulness is huge. It’s a huge opportunity.”

Cleaning robots have huge commercial value in the era of epidemic

The Links:   SX14Q003 SKM200GAL176D DISPLAY-COMPANY

Related Posts