This week we examine a new lensless camera. The opportunity to miniaturize photography has arrived. We look at a new robotic laundry that will remove the need for humans to complete this most mundane of the mundane tasks. We investigate a couple of new patents from Apple that might be used in upcoming VR and AR products. Finally we visit a dam in Tibet that is being 3D printed with minimal human labour.
Lensless Cameras
Current digital cameras require a lens to focus incoming light on an image sensor. The same way that our eyes focus light so that we can see. Lens technology has continually progressed and become smaller and smaller however there is a limit to how small they can be made due to the laws of physics. There is a limit to the size of the lens and the length to the focal point.
Researchers at the Tokyo Institute of Technology have developed a new approach which allows a lensless camera design. The lensless camera uses an image sensor and a thin mask in from of the sensor that encodes information from the scene being photographed. A mathematical reconstruction then produces a detailed image.
The system relies on a novel deep learning system to produce the reconstructed images. This will allow the size of cameras to become ultra miniature bringing new applications that we can only currently imagine. We will have to assume that cameras are everywhere.
Apple Glasses Patents
It has long been rumored that Apple is working on new Virtual Reality Glasses. Two recent patents point towards some interesting developments towards those glasses.
People who wear corrective vision glasses have difficulty in using some VR headsets due to the need to wear their prescription lens whilst using the device. This makes the the use of the headsets uncomfortable and cumbersome. Apple has applied for a patent that will build vision correction into the headset itself.
The system which Apple calls “Tunable and Foveated Lens System” uses a stack of lenses for each eye to perform vision correction. The patent mentioned liquid and non-liquid crystal adjustable lens. They are likely to be used in combination. Power is applied to the lenses which would modify how the material permits light to pass through. The voltage setup and 3 to 6 layers of lens would allow individual users to correct the lens to their own requirements.
A second patent helps to automatically clean the lenses of the VR device. When dust or other debris clusters on the lenses they can block pixels which can result in very different images than intended. This problem gets worse at the device ages. The patent describes a way that the device will vibrate and trap the dust and debris inside a mini container automatically.
A cleaning cloth will still be required for fingerprints and smudges. There are rumors that Apple will release their VR headset later this year or early next. The next project will be augmented reality glasses (picture above) however it is likely some time before they come to market.
Robotic Laundry
Researchers at University of Bologna and Electrolux have developed a new robotic system to assist with doing the laundry. They have trained the robot to insert items into a washing machine and pick them out again once a washing cycle is complete.
The robotic system detects clothes that need to be washed, picks them up, puts them into the washing machine, interprets the settings on the washing machine’s display and adjusting the settings to the desired cycle. The team needed to develop several algorithms for the robot to function. A pointcloud based perception algorithm was developed to help the robot to grasp and manipulate items. A deep learning algorithm was developed to allow the robot to read and interpret washing machine displays and an interpretation algorithm was developed to fully automate the laundry operation.
The goal is to replace human operators at laundry cleaning stores and services with fully robotic mobile arms. This is not an easy task as robots struggle with highly deformable objects (such as clothes). The future steps include testing the system in real world settings.
3D Printed Dam
China is using an AI driven 3D printer to construct a 180 meter high dam on the Tibetan Plateau. The project will use unmanned excavators, trucks, bulldozers, pavers and rollers to build the project. When finished in 2024 the hydropower plant attached to the dam will provide 5 billion kilowatt hours of electricity each year.
3D printing and dam construction share several similar processes. A team from Tsinghua University have been developing the idea for the past 10 years. Initially the 3D printing was designed to reduce the amount of waste produced in the project. Additionally Chinese civil engineers had used AI to coordinate the construction of the Baihetan dam (the world’s second largest).
The team sliced the dam design into layers. The AI was then used to assign a team of robotic equipment to build the dam one layer at a time. Unmanned excavators can identify and load materials from stockpile yards into fleets of automated trucks. The AI then instructs the trucks to deliver the material to where it is needed at the right time for robotic bulldozers and pavers to construct a layer of the dam. Automated rollers then press the added layer ensuring that it is tight and firm.
Deep reinforcement algorithms allow the machinery to recognize nearly all objects on the site, deal with changes in the environment and perform various tasks flexibly. The machines can work continuously, 24 hours a day. The technology may also be suitable for use in the construction of airports and roads.
Paying it Forward
If you have a start-up or know of a start-up that has a product ready for market please let me know. I would be happy to have a look and feature the startup in this newsletter. Also if any startups need introductions please get in touch and I will help where I can.
If you have any questions or comments please comment below.
I would also appreciate it if you could forward this newsletter to anyone that you think might be interested.
Till next week.