September 9, 2016
Steve Cousins is the CEO of Savioke, which is creating robots for services industries. He is passionate about building and deploying robotic technology in places where people live and work.
Steve Cousins of Savioke describes advancements in navigation that are enabling robots to safely work directly with people in unpredictable environments.
PwC: Steve, you have been involved in robotics for a long time. Can you briefly share some highlights of your journey?
Steve Cousins: My interest and experience with robotics goes back to when I was about 3 years old and my sister got a remote control doll. It was a wired remote, and I was so jealous that she had such a cool toy. When I was an undergraduate in the 1980s, I worked with a robot arm and was a little bit frustrated that I couldn’t make it hold a pen and write letters. At that time, it was difficult to make robots that could do useful work, but I was fascinated by them.
When I was at Stanford, we worked on a mobile robot similar to what we have at Savioke except that it had a much more primitive sensor set. To make that robot go from point A to point B on a single floor of a building was a really challenging problem. The indoor navigation problem wasn’t solved in an academic sense and wasn’t really demonstrated until around 2000.
Since then, there have been a lot of changes in robotics. Industrial robots have been around for more than 50 years now, and they’re mostly cool automation in a controlled environment. During the last 10 to 15 years, the navigation problem has been solved and we’ve seen interesting little robots become part of many people’s lives. There are about 12 million Roomba robots in the world. We have learned that people are willing to bring robots into their homes and have them vacuum, clean pools, or entertain. It’s something that is fascinating.
PwC: So where is robotics headed? How will the future be different?
Steve Cousins: We are getting to the level where robots can do much more sophisticated jobs in dynamic environments. We now see robots moving around in human spaces, and they aren’t just following lines. They don’t just stop when you get in front of them, but actually will make another plan to move around you or other obstacles. Our product Relay does that.
The robotics community has figured out how to make robots act much more intelligently. They’re not smart; they’re not conscious. But they are building a model of their world, rather than being given a fixed model, as in industrial robotics. They’re exploring their worlds, building a model, dealing with changes, and responding appropriately.
Another key achievement of the past decade or so is the ability to navigate in real time. I remember there was a robot in the 1970s at SRI called Shaky; it was one of the first mobile robots. It had stereo vision, batteries, and motors—the most advanced technology of the day. But it moved very slowly, because it took a picture and then took a long time to analyze it.
“Real-time ability is not just about the compute power, although that is a big piece of it. It’s also the fact that the algorithms for moving and, more importantly, approximation algorithms have become a lot better.”
Real-time ability is not just about the compute power, although that is a big piece of it. It’s also the fact that the algorithms for moving and, more importantly, approximation algorithms have become a lot better. A robot doesn’t need the absolute best path. Just provide a path that works, but give it in the time that the robot needs it, instead of spending two days to get the very absolute best path.
People have really understood such problems on a theoretical level, and now that theory is filtering down into pragmatic and practical algorithms that can be implemented and eventually enable robots to work in real time.
PwC: Can you tell us about Savioke and your ambitions there?
Steve Cousins: Sure. We are creating autonomous robot helpers for the services industry. We aim to improve the lives of people by developing and deploying robotic technology in human environments. Our first product, Relay, is a mobile delivery robot for the hospitality industry. It delivers things, such as snacks and amenities, securely and efficiently to hotel guests, so hotel staff can focus on other guests’ needs. We built Relay on top of ROS, the open source Robot Operating System. Using ROS, we were able to build both the hardware and the software and put a robot in the field in 10 months. That is amazing.
The core technology we are developing is the ability to navigate in human spaces. In a nutshell, that means don’t hit anybody, don’t hit anything, don’t get lost, and move gracefully because you don’t want to scare people. Also, share the world with the humans—so, for example, successfully ride the elevator alone or with people. These are the key pieces of navigation we’ve developed for getting around in large human spaces.
PwC: You previously were CEO of Willow Garage, which was somewhat of an academic organization. Now you are running a startup. How are your priorities different?
Steve Cousins: At Willow Garage, we were funded to change the world and try to do something radical. The focus was more academic and less commercial, something you can’t really afford to do at a startup. We developed PR2, a humanoid platform for mobile manipulation. It was designed to allow researchers to do a variety of things and to advance the science of robotics. It was big and complex to enable a wide range of use cases.
“Our focus has been to build the simplest robot needed to perform the task. We can always come back to the core and build something more sophisticated on top of it.”
Our focus at Savioke is different. At some point, my thinking—and the thinking of others who came with me to Savioke—shifted: What’s the simplest robot we can build that would make a difference in the world? That is what we are trying to do at Savioke. As we developed Relay, we argued about every single motor: Is it necessary? Is there a simpler way to do this? We focused intensely on the human-robot interaction: Can it be simpler?
We had learned a lot working with PR2. PR2 was cool, but it was also huge. Even though it was designed to be safe, when it is 500 pounds and 5-1/2 feet tall, it can be intimidating to people. So we decided to make a robot that’s relatively lightweight, shorter than most people, and designed to look friendly.
Our focus has been to build the simplest robot needed to perform the task. We can always come back to the core and build something more sophisticated on top of it. We must walk before we can run.
PwC: How should businesses think about the value that is possible from robots?
Steve Cousins: If robots are going to move things around in the world, and if moving things around in the world is currently done by people, then you must determine the price point at which you could be competitive and actually provide value.
It doesn’t necessarily mean you’ll replace people. Factories present a lot of good examples, where a factory owner decides to bring in robots to perform some of the mundane, dangerous, or tedious tasks or tasks that might result in repetitive stress injuries. The people get trained to maintain the robots and do other tasks. That has been a recurring story from a lot of the factory automation opportunities. The overall factory throughput goes up, the salaries of the people in the factory go up, and nobody loses their job.
The same thing happens when you bring robots into hotels or anywhere else. Does anybody write on their resume that they’re skilled at walking down hallways without bumping into things and they know how to ride elevators? Nobody writes that on their resume. Those are such mundane skills that you don’t even bother to mention them. That’s the skill level our robots have.
PwC: Can businesses get enough benefit from robots while using them to perform the mundane tasks?
Steve Cousins: Indeed, the benefits are at two levels. The first benefit is a direct gain in efficiency. In the hotel context, if any staff member spends 20 minutes delivering something, most of that time—15 to 18 minutes—is walking up and down long hallways and riding elevators. So to the extent that you can automate that mundane part of the work, you can let that person spend more time helping guests in the lobby and checking in guests faster. Workers can move into a higher skilled job and not spend so much time doing the lower skilled job and being paid accordingly.
The other benefit is that because robots can drive down the cost of deliveries dramatically, like one or two orders of magnitude cheaper, the hotel can ask: What else could I deliver? What other services could I offer? We found that hotels are cutting some services, because the cost of offering those services is too high. We’re providing a way for hotels to offer old and new services, thereby bringing back and raising their service levels. A hotel can become more profitable, just like the factory that can do more work with greater automation.
PwC: What lessons you have learned from your customers so far?
Steve Cousins: The biggest lesson we have learned is that people like robots. In fact, people kind of demand robots to make the delivery. I don’t think it’s just novelty. I think it’s also that when somebody comes to the door, for some people there is awkwardness: Do I need to tip the person? I asked for a toothbrush. How much should I tip for a toothbrush that is worth 50 cents? There’s also awkwardness around a stranger coming to their door. A robot opens up new possibilities where people will be more accepting. That’s been a surprise for us.
PwC: What are some challenges that the robotics industry is facing?
Steve Cousins: One challenge is that the volumes of robots are not big enough to allow us to develop new sensors just for robotics. This is starting to happen now as people see more money flowing into robotics. The most popular sensor in robots today is the Kinect sensor from Microsoft, which came out of video games. The other sensor that’s really popular is Lidar, which came out of industrial automation. We can repurpose these technologies to do navigation well.
Sometimes I say that robotics is the poor stepchild of every other industry at this point in history. I don’t know if that’s always going to be the case. I hope not. We’ll get our volumes up.
PwC: Do you think enough growth will occur in the next five years to get over that volume issue?
Steve Cousins: Yes, we’re getting there. As an industry, I think we’ll be in the millions in 5 to 10 years. I hope we can do that as a company.
There’s a real pain point that robots address. As an individual, I already buy things that we don’t call robots, but they’re robotic in nature, because they automate some task. They’re called appliances. People buy those all the time, whatever they have space for, because they help. Appliances free us from scrubbing our laundry, for example: just throw it into a machine and push the button, and move it to the dryer and it gets out.
There are real pain points in a business context as well. If you look at automatic restaurants, what we’re doing in hotels, and delivery robots in hospitals, these are all compelling use cases. Usually, if an institution has decided to adopt a technology, it means there’s some perceived significant cost savings and they’ll be more competitive with the technology.