Service Robotics Case Studies - Silicon Valley Robotics

CASE STUDYFetch Robotics Core TeamInterview with Melonee Wise,CEO Fetch Robotics edited for clarityWhat’s the current value proposition for FetchRobotics?We see the largest opportunity within the logisticsand material handling industry. E-commerce andan on-demand economy are putting tremendouspressure on inbound logistics and fulfillment. Ifyou look at the top issues for logistics managers,there are several different types of issues thatthey’re looking to tackle. One is throughput,obviously. Another one is insight into the dayto-day happenings of the warehouse. Logisticsmanagers are making decisions based on datathat is sometimes almost a month old in manycases. Lastly, being able to add technology intoan aging warehouse. Those three things are whatwe’re trying to capture at Fetch.Fetch RoboticsFetch Robotics builds robot systems for the logistics industry. The companywas founded in 2014 and is headquartered in San Jose, CA. Unveiled in April 2015,the Fetch Robotics system is comprised of a mobile base (called Freight) and anadvanced mobile manipulator (called Fetch). Fetch and Freight use a chargingdock for autonomous continuous operations, allowing the robots to charge whenneeded and then continue on with their tasks. In addition, the system includesaccompanying software to support the robots and integrate with the warehouseenvironment. The robots are designed to work independently alongside humanworkers, performing repetitive tasks such as warehouse delivery, pick and pack,and more. Fetch was awarded a GameChanger award by Robotics BusinessReview for Best Industrial Productivity Solution.How did you identify logistics managers asyour alpha customers out of the whole range ofservice robotics?We first decided to focus on the logistics marketas there is a great deal of supporting evidence thatthe market needs to turn to automation.One obvious indicator is that the market ischanging. There’s a lot more demand for instantdelivery, or express delivery, same day delivery,same hour delivery. Businesses somehow have toget goods from a warehouse to a person’s handsin a shorter amount of time. The other factor isthe current job gap. There are about 600,000unfilled jobs in this industry and most peopleestimate that’s only going to get worse.Second, as warehouses add more automation,they’re not going to want to knock down awarehouse every time they do so. That meansthere’s a market opportunity for installing “in-placeautomation.”10 2015 Silicon Valley Robotics. All rights reserved.Melonee with Fetch and FreightHow many trials does Fetch have underway?Right now we are working on three differentcommercial pilots, all at different stages. We can’tsay much about them just yet but we are in theprocess of doing commercial deployments. We’vealso sold several robots into research, which isalso a means for us to hire as we grow. If we getrobots into research labs then students are alreadyfamiliar with our robots. So that’s not so much asales channel as a recruitment channel.Looking at your commercial customers,what would you say are the similarities anddifferences?Some of our current customers just want to usethe software and the robots as we’ve imaginedthem in our videos. But others want us to changeit up a little bit and do point-to-point deliveryor have robots going between people insteadof robots following people. There are a lot ofoptions. Some of it is nonrecurring engineeringwork and some of it isn’t.Lastly, the logistics market in general workswith such thin margins. They have to increaseproductivity or risk losing margin.fetch robotics11

Are you offering robots as a service, completelyor partially? (What is your business model?)talking about a robot that just follows you, it’smuch simpler to understand.Right now we aren’t specifically offering robotsas a service. We are doing engineering work tocustomize the robots. In terms of our businessmodel, we’re still coming to a final solution onour pricing strategy. We have already seen thatsome potential customers prefer the capital costsof buying the robots, where others would preferthem as a service.At this point, we’re not worried. We’re goingto work with our customers to help themunderstand the capabilities of our robots and webelieve that we’ll get more traction in the future.What would a typical deployment of Fetch andFreight look like in the future? How many unitsof each and how would it integrate with existingsystems?Right now follow-pick is our most popular kind ofproduct, wherein our Freight robot simply followsan individual in a warehouse. Our customersand prospect are still looking at the problem.Do they want better transparency into theirwarehouse? Do they want better throughput?How much more throughput do they want?How big is the warehouse? How many peopledo they have picking. So there are a lot of factorsdetermining what is a standard system, meaningthat essentially there is no standard system. We’realso trying to find ways to talk about it like robotsper square foot or robots per employee but we’renot there yet.We seem to find it easier to imagine robots thataugment human workers, like Freight, perhapsCan you describe Fetch and Freight, what theydo, what they cost and what the ROI is?Fetch and Freight are a pair of robots. Whenthey’re working autonomously they’re most oftenworking as a team. Fetch is a mobile manipulationrobot. It has an arm and can pick up objectsweighing up to 6 kilograms. Fetch can navigateautonomously, search out objects on shelves andpick them up. Freight is a mobile robot. It’s meantto do most of the freighting or transporting in awarehouse because fifty percent of a picking taskis transport.We decided that when you look at the cost ofmaking Fetch vs. Freight, it’s far more inexpensiveto make Freight. Fetch is more sophisticated,more like a person. Like a person, when Fetchisn’t picking then it’s under-utilizing its capabilities.That’s why we developed this buddy pair. In termsof cost, we’re really not talking about that yet. Thecost of the system is extremely dependent on thesize of the system, the volume of the order, howmany warehouses it’s going into and all of that. Allof those things factor into pricing our robots.Fetch is too big a step?What is the ROI for a warehouse?We’re positive that it’s less than two years butdon’t yet have data yet to effectively support that.And what about the difference in timing?Whether or not you are serving legacy hardwareor not?Legacy is a big problem for the industry.What so far is staff response to it?We haven’t really gotten past novelty yet, so we’llsee.What are some of the issues you’ve faced indeployment?Most of the issues we’ve faced involve migratingfrom existing back-end systems. Warehousesystems are a hodgepodge of differenttechnologies.Do you have specific lessons from thesedeployments? Has that lead to changes in whatyou’re doing?Not yet. We haven’t been deployed for verylong, so we’re trying to not make any preliminaryconclusions.What is the competitive market? There is verylittle that is a mobile manipulator but there arequite a range of mobile solutions, presumably inwarehousing as well as other service areas? Canyou describe how you differ?This is really a hard question because its moreabout the software than the robots. The hardwareis very similar. We know that we are more pricecompetitive, but it really comes down to thesoftware that we’re using.What are the environmental factors thatcontribute to Fetch that might not have beenpossible five years ago?ROS, definitely. It certainly helped that manyof the core members here at Fetch were alsoon the ROS core team. Sensor technology hascome quite a ways. Also, the way we think aboutdoing things has changed significantly becausecomputational power has finally caught up withsome of the algorithms. A really great exampleof this is particle filters. Our use of particle filtersfor localization has been limited so far, to usinga small number of particles, because of thecomputing power needed to do the algorithmfor a certain amount of particles. The momentyou wanted to add thousands and thousands ofparticles it wasn’t practical. But now computingpower has gotten a lot more powerful and sosome of the things we thought about previouslybut couldn’t implement because of the limitedcomputational power, well now we’ve changedour thinking significantly. We’re starting to usethat computational power more effectively to dointeresting more novel things, which improvesthings like better localization.Freight robot, docked and undockedIt’s very possible that Fetch is going to be anincremental step because of the discomfort somebusinesses have regarding automation. Whenyou look at Fetch versus Freight, Freight is a lotmore understandable and businesses have abetter understanding of how it translates directlyto productivity. When they look at Fetch theywonder why it looks that way, how many picksit can do, how accurate it is, etc. When you’re12 2015 Silicon Valley Robotics. All rights reserved.fetch robotics13

What new areas will that open up for robotics?Let’s assume that we’re reaching a point whereour technology is able to meet the needs ofservice robots like Fetch and Freight? Where doyou see future potential?I think there are market opportunities, specificallyin elder care. Everyone sees that as the next bigthing. The challenge there is whether we’re goingto be able to get around the safety issue. Thereare a lot of edge cases and I just don’t know if wecan do it. We also haven’t done the same legworkthat we’ve done in logistics, so it’s hard to naildown where we want to go next.Fetch robot, docked and undockedWe’re going to look at light manufacturing: robotsactually doing the assembly or the value-addtasks. And that’s actually more specific to Fetchrather than Freight, although perhaps Freight canvalue-add too. I expect there are some value-addservices where the robots can probably extendinto after logistics.There’s one key thing, people are not necessarilygreat at walking compared to everything else wedo well—excluding the health benefits—walking isnot a high value activity for people.Finally, are there messages that you’d like to getout about Fetch and Freight?We’re trying to make it clear to the world thatFetch Robotics is providing solutions—bothcollaborative and autonomous—for the logisticsspace. We don’t want people to think that ourrobots just do x and y. We’re hoping to engagepeople in the idea that we can create newapplications and capabilities for these robots, toextend them beyond what our initial vision hasbeen.CEO Melonee WiseO’Reilly AlphaTech VenturesOATV, or O’Reilly AlphaTech Ventures, is a seed stage investment firm basedin San Francisco with a track record of backing robotics startups in emergingareas. As well as investing in Fetch Robotics, some of their other hardwareinvestments include 3D Robotics, Planet Labs, Misfit Wearables, Littlebits and SightMachine. OATV typically invest between 250,000 and 2 million into startups at acritical early stage of development. This pre-revenue runway helps startups refinetheir prototypes, determine product/market fit, and achieve strong follow onrounds. OATV invests early, typically before market categories are well defined.Interview with OATV Principal Roger Chen edited for clarityWhat’s OATV’s investment thesis andhow does that make you look at roboticscompanies?more capital into those companies and takemore of an all-in approach. Our personalphilosophy and style is to try to work moreclosely with the companies we invest in, and itbecomes hard and unwieldy to do that if youinvest in too many companies within one year.We like to look on the edge. If you take alook at our portfolio, there’s a wide array ofcompanies in different categories. You haveanything from consumer internet companieslike Foursquare to satellite companies likePlanet Labs to drone companies like 3Drobotics and logistics companies like FetchRobotics. We made each investment whenwe sensed the emergence of a new category.When we think a “thing” is going to becomea “thing,” we try to find companies andentrepreneurs in those categories very earlyon and back them just before the categoriesare really created. Before 3D Robotics, droneswasn’t really much of a category. And neitherwas space before Planet Labs and otherpioneering companies like SkyBox.When it comes to robotics, we are seeing alot of interesting things happen. There hasbeen a confluence of technologies enablingnew forms of robotics, from innovationsin actuators enabling compliance andcollaborative robotics to innovations insensors and software.ROS, or Robot Operating System, onlyemerged these last few years, and beforethat, software development for robotics wasexceedingly difficult. The advent of opensource communities and platforms like ROShas really catalyzed the field. So those aresome of the enablers on the technology side.Our focus has been on that strategy applied atthe seed stage. We’re a little bit different thanother seed firms in that we invest in fewercompanies, about six a year. We concentrate14 2015 Silicon Valley Robotics. All rights reserved.ANALYSISWe’ll be able to use more computationallyintensive algorithms that might produce bettervisual or navigational results, more naturalbehaviors, more clever solutions. From myperspective, it’s one of these things wheretechnology seems to be time capsuled by thetechnology that was available at the time it wascreated. Robotics is specifically susceptible to thatbecause it’s so dependent on new sensors andincreased computational power. Until everythinggets sufficiently fast or powerful you’re alwaysgoing to have this legacy problem where we didthis because of x, y or z limitation at the time.Logistics is pretty broad? At the moment you’retargeting e-commerce warehouses, are thereother applications in logistics?There’s also a lot happening on the marketside, and market pull is just as important astechnology. Just to give one example, let’sOATV15

take a look at e-commerce and what’s happeningwith consumers. They want things faster, cheaper,and personalized, and this just creates so muchpressure on a lot of these manufacturing, supplychain, and logistics companies.At the same time, there are macro trends withinthe labor economy, as baby boomers startgetting older and labor supply is expected todrop significantly. The confluence of all thesefactors puts logistics companies in a tough spot.People tend to forget that someone somewherestill needs to make, package, and ship thingsas part of e-commerce’s backend. Suddenly,automation and robots make a lot of sense andare economical.That’s how we see the robotics industry, andwhile that example is specific to the logisticsindustry, I think there are a lot of industries whereautomation and robotics are going to come intoplay in similar ways. It’s going to be a collisionbetween technology enablers on one side andintense emerging market demand on the other.I can give you another statistic on the labor side. Alot of people have concerns about how roboticsand automation will disrupt labor, which I think isvalid and true to a certain extent. But I think youalso have to be nuanced about it because if youactually look at the manufacturing and materialshandling industry, particularly in the US, there is ahuge job gap of 600,000 people because thereis not enough sufficiently skilled labor to executeon the work to be done. That then presents anopportunity for robotics and automation to comein and fill that gap.These are the powerful forces we see drivingrobotics: the really intense demand for logisticsfulfillment, and simultaneously a lack of people todo all the work at an economical cost.What other areas of the commerce valuechain can robotics, smart automation and AIpotentially improve?I’ll talk a little more about market pull. Dependingon country, online sales make up somewherearound 10% of overall retail. You can see howmassive companies like Amazon and Alibabaare, and that 10% online share of retail will onlycontinue to grow. It speaks to the growingvolume of work that has to be fulfilled on thebackend of e-commerce.There’s a ton of room for cost reduction viaautomation, but it’s not necessarily with justphysical automation. There are software solutionsas well that can make supply chains a lot moreproductive. For example, our portfolio companyFetch Robotics is tackling the logistics problemby streamlining operations in factories andwarehouses with a mobile robotics platform.However, Fetch will be just as much about itsfuture operations management software andapplication data as its material handling robots.I think a lot of people, especially consumers,don’t see how much work has to go into fulfillingthose online orders. There is a box that has tobe moved. There is something that has to bepacked. There is something that has to shippedand transported. That’s kind of shielded awayfrom consumers’ eyes. But it all has to be done,and it’s becoming more and more challenging forAt some point, all the goods that a company likeFetch moves around will need to be packagedinto containers to be shipped all around theworld. Another OATV company called Havenis creating a marketplace for streamlining howcontainer shipping is booked. If you think aboutit, it’s rather ridiculous that people still have tocall one another and use manual paper-basedCould you expand on some of the trendsenabling robotics and automation in logistics?16logistics companies to fulfill all these operationseconomically. 2015 Silicon Valley Robotics. All rights reserved.processes to mix and match which containersshould go on which ship. It’s just very inefficient,and it hurts business by not maximizing fulfillmentof shipping capacity. This is a case where relativelysimple automation through purely software and aweb application can go a very long way in drivingup productivity in the supply chain.The take home message here is that when I thinkabout automation for improving supply chains, it’snot necessarily just robots with arms that movearound and pick things up. It’s as much about thesoftware as it is about the hardware.What are some examples of OATV portfoliocompanies, perhaps not robotics companies butwhere the lessons can be applied to robotics?I just talked about Haven a little bit – it’s essentiallyan online marketplace for more efficiently fillingcapacity on ships for shipping things.There are a couple other OATV companiesthat come to mind. One is Sight Machine. It’sa data platform company. They aggregatedata streams across the manufacturing floor,perform analytics on them, and offer a frontenddashboard for customers to understand exactlywhat’s happening along their manufacturinglines. That has a ton of value because that kindof intelligence is what will allow decision makersoverseeing operations to keep things up andrunning efficiently.Another example is Riffyn. They are similar toSight Machine in that they are also a data platformthat aggregates data streams, but their focusis on the R&D lab for life science companies.Currently 70-90% of R&D results in the lifesciences are not reproducible. If you are a pharmacompany, imagine the egregious amounts ofR&D money wasted due to poor process control.Companies like Riffyn perform data automationto give science-driven enterprises control of theirprocesses again. They automate data collection,root cause analysis and continuous deploymentof improved process designs to drive upproductivity for the R&D pipeline.Both Sight Machine and Riffyn automateworkflows to enable superior operationalintelligence, flexibility, and performance to driveup productivity. While neither company is arobotics company in the traditional sense, thatexact value proposition very much applies torobotics as well.How is robotics today different from robotics inthe past?Once upon a time, robotics was aboutstationary, highly repetitive, high performance,and generally expensive automation. Industrialrobot arms would repeat the same action againand again with extreme precision. We’re not asexcited about those applications. We think a lotmore about flexible robotics. We think aboutprogr

Silicon Valley Robotics was launched in January 2010 by a coalition of 30 robotics companies that included founding board members Adept, Bosch, SRI International and Willow Garage. Rich Mahoney, Director of Robotics at SRI International is the Founding President of the association. Andra Keay is the Managing Director, in charge of daily operation.