The IROS 2014 Industry Forum brought together entrepreneurs, researchers, venture capitalists and funding agencies from North America, Europe, Asia and Australia to shed perspective on the above questions and engage in open dialogue about some of the pitfalls of commercializing robotics technology. Here are the 10 best quotes from the event …

1. “Testing nothing and testing everything are equally lazy.”
– Ryan Gariepy, Clearpath Robotics

Gariepy refers to the extremes some startups take before launching a new product. He instead suggests evaluating the technologies, processes and people that can be reused efficiently, then taking a decisive step to engage your customer.

2. “Competition is not other companies or technologies, it’s the status quo.”
– Ryan Gariepy, Clearpath Robotics

Academics, consumed with dissecting the research of predecessors, colleagues and other institutions, then publishing novel work, naturally look to see what others of doing in the field to gauge their own progress. To academics looking to commercialize research, Gariepy says it’s important to know what’s out there but also to run your own race. When launching a new product, you need to focus on market first, then technology.

3. “Cheaply eliminate bad business models fast.” 
– Lee Redden, Blue River Technology

Before building their first product, Redden and team hired five part-time interns to talk with potential customers, and eliminated five poor business models in less than a month.  CarrotBot, a weeder for carrot farms, came about after talking with an IT procurement manager at a carrot farm who had struggled with finding the right technology for their particular problem. Redden also notes that if you want honest feedback from potential customers, make them feel comfortable by sharing the flaws of your product upfront and don’t show up with a perfect prototype – most people are hard-wired to protect the feelings of those they interact with, so you have to disarm them to get honest answers. “Cheaply eliminate bad business models fast.” – Lee Redden, Blue River Technology 

4. “Sometimes what the company wants and what the community wants are not perfectly aligned.”  
– Brandon Basso, 3D Robotics

Balancing of the vision of the company with the wants of an open-source community can be tricky. The 3D Robotics (3DR) community wants breadth – more hardware, sensor and platform support – whereas the company wants depth to support the community while developing native system apps and more hardware solutions. Challenges arise when a growing company has to execute on set product and revenue objectives defined by management (and in 3DR’s case, venture capital stakeholders) while integrating the community’s feedback into their product suite. Long-term, this relationship creates a virtuous cycle. The community provides 3DR with an on-demand workforce, tech support via developer list, rapid exploration of new opportunities and a huge beta test group. 3DR gives back to the community through cheaper and faster hardware, expertise, capital, support and worldwide distribution.

5. “The robot revolution will be an evolution.”
– Christopher Parlitz, SCHUNK

According to Parlitz, the integration of components is the new challenge in robotics. Without standards in place for mechanical and electrical components or transfer of data, creating intelligent systems will take time. He talks of Programmable Logic Controllers (PLCs) and ROS as two solutions for integrating hardware components, each with its list of pros and cons. As new integrated intelligent solutions start working robustly, they will no longer be considered “robots”, making the progression much more of an evolution than a revolution. “The robot revolution will be an evolution.” – Christopher Parlitz, @SCHUNKInc 

6. “If I lose control of my startup, my invention won’t wind up at a fire sale.” 
– Michael Peshkin, Northwestern University

IP. IP. IP. Intellectual property dominated this year’s industry discussion. Each university’s tech transfer office has its own rules – some grant both faculty and student inventors full ownership, others to students but not to faculty, and then there are jointly-owned inventions between university and industry (Renaud Champion, a venture capital partner at Robolution warns researchers against these). Emotions and the perceived value of owning intellectual property can often outweigh its real monetary and practical benefits. Peshkin speaks to this point and outlines six key reasons why he prefers Northwestern University’s technology transfer office to own his patents:

• They pay the expenses, which can cost hundreds of thousands of dollars
• If they don’t want to pay, they unambiguously turn the rights over
• Future investors shouldn’t worry about IP ownership
• They take a reasonable percent of royalties and/or equity
• They won’t ever sell a patent
• “If I lose control of my startup, my invention won’t end up at a fire sale” – Michael Peshkin, Northwestern University 

7. “If there are people with pitchforks at your door, you’re probably doing something right.” 
– Shahin Farshchi, Lux Capital

In addition to IP, regulations (or lack thereof) of certain robotics products create an uncertain environment for commercializing technology. Farshchi says focus on building a company with a clear value proposition that customers want. If regulatory bodies want to speak with you, it’s probably because you’ve taken enough market share to draw attention. And as a first mover, you’ll more likely have the opportunity to shape regulation which will put you in a favorable position going forward.

8. “Good research doesn’t necessarily mean a good product.”
– Ayanna Howard, Zyrobotics

People buy things that are useful to them and shifting focus from product specs to selling a value proposition can stumble some researchers. That’s why Howard says it’s important to be self-aware and know when it’s time to bring in outside expertise. She sees the quality of management as being just as important – if not more important – than the technology itself.

9. “If I have competition, there’s a market, if there’s a market, there’s an industry, and if there’s an industry, there’s liquidity.”
– Renaud Champion, Robolution Capital

Competition is not always a good thing, but in a nascent market for robotics products, it can be great. Competition means there’s enough market demand for several products to compete with one another and from a venture capitalist’s point of view that means there’s room for growth and potential exit opportunities.

10. “I have a dream. I have a vision. Why not me?” 
– Renaud Champion, Robolution Capital

At some point, starting a new business takes a leap of faith. Dreams can take a long time to manifest. Finding the right (or sometimes good-enough) answers to build a solid vision, a product, a team, a brand and convince customer and investors that your product can bring value to the community is tedious and was compared to a rollercoaster ride by many of entrepreneurs at the forum.  But this process, which almost always transforms the entrepreneur (and if they’re lucky, an entire industry), often starts with a simple question: “Why not me?” “I have a dream. I have a vision. Why not me?” – Renaud Champion, Robolution Capital 

Follow all of our IROS 2014 coverage here.

Tesla CEO, Elon Musk, recently announced that the car manufacturer will produce self-driving cars within three years. Nissan has announced that it will have a self-driving car available by 2020, Google has said it will do so by 2018. Over the past decade, the conversation around self-driving cars has evolved from futuristic police chase sequences in Minority Report to figuring out which auto manufacturer will be first to launch a commercially viable self-driving vehicle. Daimler AG, maker of Mercedes Benz, recently announced that an S-class sedan had completed a 62-mile journey in the streets of Germany without a driver. Audi’s self-driving car successfully navigated 156 turns of the 12-mile Hill Climb course in Colorado’s Pikes Peak. Car manufacturers see self-driving cars as a way to eliminate road deaths caused by human error, reduce traffic, and free up time spent commuting – but how do these vehicles work?

Self-driving cars in a nutshell

A self-driving car is capable of sensing its environment and navigating without human input. To accomplish this task, each vehicle is usually outfitted with a GPS unit, an inertial navigation system, and a range of sensors including laser rangefinders, radar, and video.  The vehicle uses positional information from the GPS and inertial navigation system to localize itself and sensor data to refine its position estimate as well as to build a three-dimensional image of its environment.

Data from each sensor is filtered to remove noise and often fused with other data sources to augment the original image. How the vehicle subsequently uses this data to make navigation decisions is determined by its control system.

The majority of self-driving vehicle control systems implement a deliberative architecture, meaning that they are capable of making intelligent decisions by 1) maintaining an internal map of their world and 2) using that map to find an optimal path to their destination that avoids obstacles (e.g. road structures, pedestrians and other vehicles) from a set of possible paths. Once the vehicle determines the best path to take, the decision is dissected into commands, which are fed to the vehicle’s actuators. These actuators control the vehicle’s steering, braking and throttle.

This process of localization, mapping, obstacle avoidance and path planning is repeated multiple times each second on powerful on-board processors until the vehicle reaches its destination.

The next section focuses on the technical components of each process: mapping and localization, obstacle avoidance and path planning. Although car manufacturers use different sensor suites and algorithms depending on their unique cost and operational constraints, the processes across vehicles are similar. The descriptions below most closely mirror their implementation in state-of-the-art self-driving military vehicles.

Breaking Down the Technicals

Mapping and Localization

Prior to making any navigation decisions, the vehicle must first build a map of its environment and precisely localize itself within that map. The most frequently used sensors for map building are laser rangefinders and cameras. A laser rangefinder scans the environment using swaths of laser beams and calculates the distance to nearby objects by measuring the time it takes for each laser beam to travel to the object and back. Where video from camera is ideal for extracting scene color, an advantage of laser rangefinders is that depth information is readily available to the vehicle for building a three-dimensional map. Because laser beams diverge as they travel through space, it is difficult to obtain accurate distance readings greater than 100m away using most state-of-the-art laser rangefinders, which limits the amount of reliable data that can be captured in the map. The vehicle filters and discretizes data collected from each sensor and often aggregates the information to create a comprehensive map, which can then be used for path planning.

For the vehicle to know where it is in relation to other objects in the map, it must use its GPS, inertial navigation unit, and sensors to precisely localize itself. GPS estimates can be off by many meters due to signal delays caused by changes in the atmosphere and reflections off of buildings and surrounding terrain, and inertial navigation units accumulate position errors overtime. Therefore localization algorithms will often incorporate map or sensor data previously collected from the same location to reduce uncertainty. As the vehicle moves, new positional information and sensor data are used to update the vehicle’s internal map.

Obstacle Avoidance

A vehicle’s internal map includes the current and predicted location of all static (e.g. buildings, traffic lights, stop signs) and moving (e.g. other vehicles and pedestrians) obstacles in its vicinity. Obstacles are categorized depending on how well they match up with a library of pre-determined shape and motion descriptors. The vehicle uses a probabilistic model to track the predicted future path of moving objects based on its shape and prior trajectory. For example, if a two-wheeled object is traveling at 40 mph versus 10 mph, it is most likely a motorcycle and not a bicycle and will get categorized as such by the vehicle. This process allows the vehicle to make more intelligent decisions when approaching crosswalks or busy intersections. The previous, current and predicted future locations of all obstacles in the vehicle’s vicinity are incorporated into its internal map, which the vehicle then uses to plan its path.

Path Planning

The goal of path planning is to use the information captured in the vehicle’s map to safely direct the vehicle to its destination while avoiding obstacles and following the rules of the road. Although manufacturers’ planning algorithms will be different based on their navigation objectives and sensors used, the following describes a general path planning algorithm which has been used on military ground vehicles.

This algorithm determines a rough long-range plan for the vehicle to follow while continuously refining a short-range plan (e.g. change lanes, drive forward 10m, turn right). It starts from a set of short-range paths that the vehicle would be dynamically capable of completing given its speed, direction and angular position, and removes all those that would either cross an obstacle or come too close to the predicted path of a moving one.  For example, a vehicle traveling at 50 mph would not be able to safely complete a right turn 5 meters ahead, therefore that path would be eliminated from the feasible set.  Remaining paths are evaluated based on safety, speed, and any time requirements. Once the best path has been identified, a set of throttle, brake and steering commands, are passed on to the vehicle’s on-board processors and actuators. Altogether, this process takes on average 50ms, although it can be longer or shorter depending on the amount of collected data, available processing power, and complexity of the path planning algorithm.

The process of localization, mapping, obstacle detection, and path planning is repeated until the vehicle reaches its destination.

The Road Ahead

Car manufacturers have made significant advances in the past decade towards making self-driving cars a reality; however, there still remain a number of technological barriers that manufacturers must overcome before self-driving vehicles are safe enough for road use. GPS can be unreliable, computer vision systems have limitations to understanding road scenes, and variable weather conditions can adversely affect the ability of on-board processors to adequately identify or track moving objects. Self-driving vehicles have also yet to demonstrate the same capability as human drivers in understanding and navigating unstructured environments such as construction zones and accident areas.

These barriers though are not insurmountable. The amount of road and traffic data available to these vehicles is increasing, newer range sensors are capturing more data, and the algorithms for interpreting road scenes are evolving. The transition from human-operated vehicles to fully self-driving cars will be gradual, with vehicles at first performing only a subset of driving tasks such as parking and driving in stop-and-go traffic autonomously. As the technology improves, more driving tasks can be reliably outsourced to the vehicle.

The technology for self-driving cars is not quite ready, but I am looking forward to the self-driving cars of Minority Report becoming a reality.

If you liked this article, you may also be interested in:

• Pure autonomy: Google’s new purpose-built self driving car
• Morgan Stanley reports on the economic benefits of driverless cars
• SAE defines six levels of driving automation
• Focus Series on Big Deals: What It Means to Have the Giants Investing in Robotics
• Focus Series: The state-of-the-art and the future of agricultural robotics

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Even so, the market for consumer robotics is growing.  The International Federation of Robotics estimated the 2012 market size at $1.2B with expectations to grow to $6.5B by 2016. In a panel discussion on the future of consumer robotics, Matt Fisher, founder of KumoTek Robotics said that in order to be successful robotics companies have to shift away from being gadgets and show customers a return on investment.  Paolo Pirjanian, CTO of iRobot, said that companies must focus on the core value they present to the customer.

Buzz and publicity often don’t translate into high sales numbers (Anchorman 2 anyone?) and from a company’s line of products perhaps only one or two will find mass appeal with customers. After speaking with founders and exhibitors at CES, here are the seven consumer robotic products that have the highest potential for mass customer adoption based on the strength of their value proposition to their intended market segment.  Some of these robots are still in pilot tests while others have paying customers.

Ozobot

Ozobot revealed its new gaming platform, a pocket-sized (1 inch) robot with the ability to operate on physical and digital platforms, at this year’s CES.  The intelligent robot recognizes over 1000 different digital codes, and is capable of reading different line and light colors on game boards, paper and digital smartphone and tablet screens using a small optic sensor.  Players can create their own game or download one of four Ozobot gaming apps available on iOS and Android devices. Ozobot will launch a Kickstarter campaign in January and sell the tiny robots for $60 each.

Ekso by Ekso Bionics

The Ekso is a wearable robotic exoskeleton that helps paraplegics stand and walk.  A series of built-in force and motion sensors monitor shifts in the user’s weight then translate the inputs into actions such as walk straight, stand up from a sitting position, and sit down.  The Ekso is currently being sold to rehabilitation centers and individuals for $140,000.

Jumping Sumo and MiniDrone by Parrot

Jumping Sumo is a spinning, leaping, jumping robotic toy that can be controlled via a smartphone or tablet.  Sumo jumps over 2.5 feet high and makes sharp 90 degree turns at high speeds. A front-facing center camera relays video to the app.  MiniDrone is a remote controlled drone equipped with indoor wheels to roll on the floor or on the ceiling. Parrot revealed both Jumping Sumo and MiniDrone at this year’s CES with a demo that showcased the toys’ leaps, high speed flying and smooth stabilization system. Both toys will be available later this year but no price has been revealed yet.

Budgee by Five Elements Robotics

Budgee carries your luggage, books, groceries, and other belongings and follows you wherever you go. Users wear a small transmitter which Budgee then follows within a constant range.  Budgee can carry up to 50 pounds and is equipped with bump sensors bring it to a stop in case it hits an object. Users can also control the robot using a smartphone or tablet. Five Elements is currently conducting pilot studies to finalize the design.

SoloShot2 by SoloShot

Soloshot released SoloShot2, its second generation robot cameraman device at CES this year that tracks and films subjects up to 2000 feet away. The device tracks a RF transmitter worn by the subject and automatically adjusts the mounted camera angle using a pan-tilt head that centers on the subject. SoloShot2 has 360 degree continuous tracking, 120 degree vertical tracking and can hold cameras up to 5 lbs.  The device is meant for outdoor use and retails for $399.

AIMe by Jigabot

AIMe is similar to SoloShot2 but intended for indoor use.  The device tracks an infrared beacon carried by the subject up to 150 feet and adjusts the camera’s position to keep the subject in the frame.  AIMe has a 360 degree horizontal range of motion and 120 degree vertical range.   Jigabot revealed a prototype at CES but the product will be available in June 2014 at a retail price of $299.

Paro

Paro is an interactive therapeutic robot developed by AIST in Japan for treatment of elderly care patients.  It resembles a baby harp seal and interacts with users by moving its head and legs, making sounds, and showing learned behaviors.  Paro can also recognize the direction of voice and words such as its name, greeting, and praise with audio sensors.  The interactive seal has been used by nursing facilities in Japan, Germany, and Netherlands for treatment of Alzheimer’s and dementia and has recently been cleared as a Class 2 medical device in the US.  Paro retails for $6000.

If you liked this article, you may also be interested in:

• Google acquires Nest for $3.2 billion
• Ekso Bionics goes public for $20.6 million
• Doggone it, there are no exciting consumer robots
• What do Google, Amazon, IBM, Apple, Microsoft and all the airlines and car companies want?
• Eureka! There are academic startups at CES 2014
• CES 2014 – Robotics on the Runway

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

CHIMP – CMU Highly Intelligent Mobile Platform – is one of 17 robots participating in the challenge.  The four-limbed robot has a wingspan of almost 10 feet and is designed to navigate on tank treads affixed to each limb.  For tasks that require use of its arms such as climbing a ladder or operating power tools, CHIMP can stand on its leg tread.  CHIMP also uses 360 degree sensing using a pair of laser rangefinder mounted to its head to visualize and map its surroundings, which a remote operator can use to analyze the disaster site.

In a recent press release, Tony Stentz, NREC director and leader of the Tartan Rescue Team said he has high hopes for the competition: “We’ve been on a fast track for the past year, doing detailed design and development of CHIMP at the same time as we were writing and testing its software on surrogate hardware. That’s an aggressive approach to producing a robot unlike any we have built and not without risk, but it appears to be paying off.”

And when it comes to his team’s readiness for the competition: “Even though we would love to have more time to practice with CHIMP prior to the trials, we’ve been pleased so far with its performance.”

The competition comes with high stakes – only eight of the seventeen teams will receive future DARPA funding to help prepare for the DRC finals next December.  All of the tasks in the competition require an extremely high level of mobility and dexterity by the robots. Some of these tasks include: driving a utility vehicle, going up a wall and cutting a hole, and opening and closing valves. If CMU’s performance in the 2004, 2005 and 2007 DARPA Grand Challenge and Urban Challenge serves any indication, this year’s Tartan Rescue Team will be a strong contender for one of those eight spots. We will still have to wait to observe the long-term effects of the DRC on the robotics industry. Will it be just another exhibition of expensive robotic gadgetry or will it expedite the commercialization of rescue robots as the DARPA Grand and Urban Challenges did for self-driving vehicles?