Hardwired: development steps, #RealSense from Intel, the #3dPrinting business
Posted on September 16th, 2015
09/15/2015 @WeWork, 115 West 18th Street, NY
Four speakers spoke about hardware products from cutting edge sensors to lessons learned developing hardware products.
The First speaker, Saar Yoskovitz @Augury talked about his company’s experience in lean hardware development. Augury was founded four years ago and produces vibration sensors tethered to a smart phone that analyze the mechanical performance of HVAC systems. Saar described their design and development process.
He started by noting that the technology is not that interesting in most products. The key questions to ask are
- What is the product?
- Who will buy it?
- Why will they pay for it?
He advocated an agile development process to get feedback for improvements as quickly as possible: fast iterations. Min effort to get max learning.
- Research : get out of the building, stick to the process. Determine the different players and what are their values. Identify risks. Create lean canvas showing key points.
- Minimum Viable Product – their initial hurdle was can they connect sensors to a smart phone?
- Alpha – start with smaller test sites – they gave prototypes to building technicians. Measure everything. Then approach larger customers- once you have a sense of the value
- Beta – you’ve proven the values, now start the design process. Get feedback by showing design options to customers. At the end of this step, put it in fancy bag and ship it to customers.
- Iterate – rapid prototyping, react -> fix -> improve. – Their initial connector was not robust so they first superglued the leads so they would not disconnect. Then they got better leads. Improve the process. Go with small iterations.
- Scale – validate the design, then invest into the machinery to create a production line
Be ready to adapt and improvise: “always carry pliers and superglue”.
Luke Iseman @ycombinator.com then talked about startups from the point of view of a seeder. Y has funded 940 startups with107 in the last semi-annual batch, taking 7% of the company for $120 in seed capital. Most of their startups have been for software, but they have increasing seeded hardware startups.
After warning that “the only reason to do hardware is to change the world”, Luke spoke about the important steps in that will help you get funded as a startup and help the chances of long term success: Make; show; iterate; sell; grow; tell; fail
- Make– create a prototype (otherwise it’s just an idea), e.g. bodyport – scale to measure blood pressure within 5 seconds
- Show – get feedback. TeaBot – creates customs blends of tea. Find out what customers want
- Iterate – improve the product. Nebia created a better shower head funded through kickstarter. They went through numerous iterations to find something that looks great and uses 30% less water
- Sell – Transcend Lighting – high efficiency led light to grow crops indoors. Found a niche market of buyers for their high intensity/high efficiency grow lights.
- Grow – Click and Grow – indoor herb garden selling plug-ins to the basic product. They are expanding their market by creating new products, such as a kitchen work table with lower, lighted shelves to grow herbs.
- Tell – compelling story otherwise you become a commodity – Tully sells condoms, but with a story and a style.
- Fail – Luna sells smart mattresses covers. They were initially rejected by Y, but have grown.
The key is to sell the product!
Between the second and third presentations, Bjorn Bollendorff @Panono displayed his product: a sphere containing 36 cameras to take a full set of images surrounding the location. Each camera has a resolution of 3meg and images can be manipulated on an ipad. They are starting to ship limited editions units ($1499) and will next create a $599 consumer version.
The third speaker, Mark Yahiro @ Intel/RealSense spoke about the sensors developed by Intel to give PCs/tablets/phones human-like spatial knowledge to change the way individuals interact with computers. The 3d camera creates a point-cloud that maps the locations of solid surfaces/points in the vicinity.
Sensors can monitor heart rate and increase the difficulty of a game when one’s heart rate increases. An app by ItSeez3d does real time 3d scanning. An application being developed with BodyLabs sweeps each side of the body and retains your measurements to exactly fit clothing and for health/fitness monitoring. He showed a video of the sensors guiding drones through a forest while avoiding the trees.
Mark then showed how Google’s Project Tango integrates these sensors with others to allow one to walk through a world created in Minecraft and integrate the virtual and physical worlds in an AR shooter game.
Matt Turck interviewed Peter Weijmarshausen, CEO of Shapeways which receives files and 3d prints objects which they ship to the customer. They print 200k designs per month.
The company was founded in the Netherlands and originally outsourced all production. They moved to New York and in 2010 they started doing their own 3d printing. Currently, they print 50-60% of their products in-house.
Over the past seven years they have seen printing evolve. Initially, printing was done using FDM (stratus) printers. Then home printers arrived and demand for FDM evaporated. They have also seen the printing expand to include gold & silver (for jewelry), porcelain (coffee mugs), etc. They have also led the evolution of the business by using smart methods to squeeze greater efficiency out of their current printers.
Peter also talked about the growing market providing accessories for drones. He also talked about how Hasbro works with independent designers who create accessories for Hasbro toys and the profits are split amongst Hasbro, the producer and the designer. This approach takes advantage of the low cost of initial production allows the community to start testing products and quickly determine what resonates with the market.
He sees a bright future for 3d printing since there are still many inefficiencies that will be eliminates by new technologies and greater efficiencies of scale with the increased sales of printers. Currently, raw materials are still expensive. Machines are built at too low a volume. This is no full color plastic. Printing is too slow. He also sees improvements in scanners and the software behind scanners to eliminate much of the barrier to 3d printing in contrast to the complexity of current CAD software.
Additional observations were
- Conductive ink not available yet, but coming to 3d printing in a year
- Carbon3d will speed up the printing process
- Eventually 3d printers will be able to stack individual atoms (MIT research)