New York Tech Journal
Tech news from the Big Apple

#Driverless #Trucks will come before driverless #cars

Posted on April 13th, 2017


04/12/2017 @MetroTech 6, NYU, Brooklyn, NY

Seth Clevenger – technology editor, Transport Topics News, @sethclevenger, talked about the rollout of driverless trucks. His main message was that there are many intermediate stages from adaptive cruise control (already exists in some cars) to fully autonomous operation.

Truck manufacturers are concentrating on systems that assist rather than replace drivers. These include

  1. Truck platooning – could roll out by year-end. – synchronize breaking; trucks can draft off each other for a 10% increase in efficiency. Brakes are linked, but still need drivers.( Peloton Technology plans to begin fleet trials)
  2. Connected vehicles – just starting to be regulated. (V2V, V2I). For instance, safety messages sent by each vehicle.
  3. auto docking at loading docks
  4. traffic jam assist – move forward slowly without driver assistance

Startups include: Uber/Otto, Embark, Starsky Robotics,

[One of my major concerns is the integrity of the software controlling the vehicle. A failure in software could cause accidents, however, my main concern is the potential insertion of a malicious virus as a sleeper cell within the millions of lines of code. In this case, the results could be catastrophic as all breaking and acceleration systems could be programmed to fail on a specific date in the future. At that moment, all vehicles on the road would be out of control potentially resulting in millions of accidents and thousands of deaths and injuries. Preventing such an event will require coordinating amongst suppliers and enforcement of strict software standards. The large number of suppliers makes this job especially complicated. This sleeper cell could lie dormant for years before it is activated.]

posted in:  AI, Internet of Things, Self-driving vehicle    / leave comments:   No comments yet

HardwiredNYC: #hardware startups, #drones, #VC, #AutonomousVehicles

Posted on May 12th, 2016


05/11/2016 @WeWork, 115 West 18th Street, 4th floor, NY

20160511_182040[1] 20160511_184103[1] 20160511_185103[1] 20160511_191823[1] 20160511_194128[1]

There were six presentations starting with two brief introductions to companies and their products. Three speakers talked about products and investing in hardware. Adam Jonas closed the evening by describing a roadmap for the future of cars.

In the first presentation, George Popescu@lampix talked about their product which projects a desktop image on any surface so that surface becomes a computer screen, a shared documents or piece of paper, etc. A projector shines an image on the surface making the surface a touch screen, drawing pad, etc. to view, edit or share materials.

Next, David @samlabs showed physical devices that can linked to each other using a visual interface. In this way push buttons can control lights, motors, tweets, cameras, etc. creating ways a non-programmer can prototype a hardware device.

Some configurations are monitoring the taking of medication from a medication box, squeezing a pillow to send a message, counting twitter tweets and activate motors when a hashtag is tweeted.

The product is somewhat similar to that offered by LittleBits.

Jonathan Frankel (Intercom system which connects anyone to anyone, anywhere using a tablet like a home intercom.) spoke about how to increase the odds of success as a hardware startup:

  1. Keep it simple, stupid – hardware has a lot of complexity. When possible choose off-the shelf. Otherwise costly and requires extra time; also better interoperability, supply chain, lead times, avoids unknown unknowns.
  2. Cash flow > BOM . need to manage growth as well as the financing arrangements
  3. Hire DB / sales early – crowdfunding may not work, so you need to start selling early
  4. Carpe diem – small window of opportunities. Seizing them makes the difference.
  5. Tips
    1. In-person > video > phone > email
    2. Get out more – network, network, network
    3. Put away the NDAs – being open gets you feedback and partners
    4. Who is on your mailing list? – follow-up with selected people on your email list
  6. Don’t work insanely hard – you need to have the emotional fortitude to overcome the valleys. So take some time off. Okay to mix business and pleasure.

NucleusLife elected to do a private presale (in favor of crowdfunding) since they wanted the ability to brand and control the entire experience start to finish. They also felt that their customer base was different from the early adopters

Next, Matt Turck interviewed Avidan Ross @Routeventures (seeds hardware startups). His interests are in physical products, with the emphasis on being an enabler in disrupting established businesses: they especially like low cost robotics and connectivity. They invest in only 6 to 8 deals per year so they can have lots of contact and input with each startup.

Their investments back their belief that robots are best when working in conjunction with humans

  1. Shapertools – handtools that assist the user when doing precision work
  2. Superflex – light weight clothing with actuation to augment human capabilities such as performing tasks involving standing and running.
  3. Plethera – software that works with 3-d plots (solidworks) to help you optimize the milling process.

They avoid one-off IoT products and hardware whose only advantage is lower production costs. They instead look for long term value and want to avoid the future struggle to maintain margins as technology and competition change over time. In the same vein, they want to price appropriately and don’t believe that products using Arduino’s or Raspberry Pi’s are scalable.

Design is important, but not core to IoT. Function comes before looks.

Dan Burton @ Dronebase talked about the rapidly evolving use of drones and their changing uses: real estate, mining inventory management, construction monitoring, etc. For instance, only within the past year has drone pilot become a profession.

Drone capabilities are increasing rapidly as a new generation of drones is created every 6 months. This has lead to the same dynamics as in smartphones, where retail products are often at the cutting edge lead by DJI. This means that most professional work is done with off-the shelf drones.

The systems making up a drone: software, gimbals, cameras, autopilots are all getting better exponentially. Battery technology lags.

Currently top end drones are accurate to 2 cm. One of the most promising next steps would be a light-weight Lidar system to get accuracy within 1mm.

Adam Jonas @MorganStanley gave a roadmap of how cars might evolve. He considered two dimensions:

  1. driver driven vs. autonomous
  2. owned vs car sharing

Based on these two dichotomies, he sees a rapid transition from owned-driver driven to shared-autonomous model of car usage. With this transition comes a change in point of view from number of vehicles sold to the number of miles traversed in a year. This transition will also create many economic winners and losers, but it is less clear who wins and who loses. Even with a transition from gas to electric cars, it is unclear whether the world-wide demand for gasoline increases (more mileage) or decreases (greater efficiency in energy usage).

posted in:  Drones, hardware, Hardwired NYC, Internet of Things, Self-driving vehicle, Tech Startups    / leave comments:   No comments yet

Self-Driving vehicle: #ComputerVision and #ControlSystems

Posted on July 28th, 2015


07/28/2015 @EqualSpace, 89 Market St., Newark, NJ


Parth, Praful, and Ak updated the group on their progress toward a demo at Bell Works on September 24. Last month the emphasis was on detecting things in 3-d. This time, the emphasis was on actuation on their current platform: a drive by wire, Baja go Kart. A video showed their program on controlling the brakes and steering using an Arduino serial interface. Their system will eventually control a MellowCab, 2 passenger electric vehicle.

They next talked about their work on computer vision using histogram clustering. They receive gray scale images (to be replaced eventually by an RBG input) and divide the pixels into rectangular blocks. They generate a histogram of the gray scale levels within each block and group together contiguous blocks with similar histograms (as measured by a chi2 distance measure). The groups of contiguous blocks are then used to detect the road. The system will eventually be supplemented by a depth camera which can measure to 20 meters.

Their main hardware in support of computer vision are

  1. Nvidia Jetson Tk1 – dev board. 2GB, Quad-core, supports Cuda
  2. Stereolabs Zed – stereo camera

Their code is posted on

posted in:  AI, hardware, Programming, Self-driving vehicle    / leave comments:   No comments yet