Home News Actuator Spending two days in the Bay Area

Spending two days in the Bay Area

167
0
Spending two days in the Bay Area

The return of rain in Northern California has had both positive and negative effects. Many people have experienced devastating floods, but typically, it seems that when it comes to the weather this new era we live in is extreme in nature.

Definitely, it is much better than the scorching 105° heat which caused rolling blackouts in Cupertino during the summer.

It’s a harsh outlook, however it is also the most truthful.

Organizations have various ideas about research. When we ordinarily consider research and development, it is work that fits in with a company’s plan. Those engineers will often work on one or two product updates in the near future. Bigger companies, with more capital, can accept a more expansive definition of the research their organization covers.

Alphabet X stands out as a notable example of this approach. In the 13 years since it was founded, this ‘moonshot factory’ has unleashed a slew of pioneering ideas, including Waymo, Google Glass, Loon, Wing, and more recently robotics companies Intrinsic and Mineral. Each individual may have varying opinions on the success of these projects, though it is clear that Alphabet X provides an enabling environment for daring and oftentimes unconventional risks not achievable in traditional corporate settings. We usually consider X to be more of an internal accelerator. Not all teams end up as startups, but it looks like that’s the plan.

I’m currently giving a lot of thought to the Boston Dynamics AI Institute, particularly as I’m set to have a chat with its executive director Marc Raibert tomorrow at the TechCrunch City Spotlight: Boston event.

The picture used belongs to TechCrunch.

Here is a brief overview of what Raibert discusses about the organization’s approach to turning research into products:

We have created a plan with several approaches. We can complete spinouts. Some regard spinouts as a way to achieve financial gain. From my perspective, spinouts are a way to safeguard the institute from any potential products.

The inception of the Institute has come about directly due to the actions of Hyundai and Boston Dynamics to turn their research commercially applicable. Even though it has only recently occurred, their work is independent from the development of Boston Dynamics. It is a captivating notion that spinouts could result from this.

Clearly, TRI is a unique entity, however in comparison to those two strategies, I would categorize it in the middle. There is a lot of research being done towards trying to improve aspects of Toyota’s automobiles such as autonomy and electric vehicles. Additionally, the teams have a significant impact over the direction of their research. This brings up the concern of meeting milestones especially in the current climate where advancements in technology are taking place and layoffs are occurring in the tech industry. The criteria for gauging accomplishment is generally dissimilar to what would be seen at a college or university.

I spoke to TRI SVP Max Bajracharya last week about the successes and struggles of their robotics initiatives. Here’s some of what we discussed and a link to the full interview:

Image courtesy of Brian Heater

TC: How do you evaluate accomplishments? What is an indicator of success for your team?

When we shifted from our homes to the supermarket, we could notice the difference in speed of improvement and its clear definition. Once in the grocery store, it became easier to observe how our system functioned and what the root of the problems were. This enabled us to concentrate our efforts on finding solutions. When we journeyed to both places, the contrast was more visible. At Toyota’s logistics and manufacturing facilities, we were presented with the grocery shopping challenge, but with a slight modification. Instead of the items needing to be purchased being groceries, they are all different parts of a distribution center.

Despite what 1,000 people have told you, constructing home robots is very difficult. Nevertheless, you decide to try it for yourself and make all the same mistakes that everyone else did.

I am definitely to blame just as much as everyone else. We become excited when our GPUs are improved and then become overly confident in believing that machine learning will make our things much easier than we initially anticipated.

At some point, something will be the deciding factor.

I don’t think there is a quick fix solution for this like there is with automated driving. Instead, it is going to be a slow and steady process where small successes are achieved over time. This is why it’s important to have a roadmap of shorter milestones that provide small wins to keep progress going towards the ultimate vision.

What is the procedure for making any of these technologies into an actual product?

We have been attempting to answer this for ourselves. It appears we have an idea of what needs to be done. It appears that I didn’t think it would be this complicated at first, presuming we could pass the technology on to Toyota or another organisation. We have discovered, though, that no matter who ultimately has this responsibility – a company, a business, a startup, or a division of Toyota – no entity like this appears to exist. We are attempting to generate a method and I assume that accurately summarizes TRI-AD as well. It was established to shift the automated driving analysis that we were engaging in into something that was more sensible. We have comparable troubles with robotics, as well as with a lot of the innovative technologies that we utilize.

I suggest you read the entire interview to gain more understanding of the reality of home robots and why we strongly desire to find out what exactly the challenges are. We were fortunate to have a straightforward discussion, something we do not come across regularly. It was like this:

We had trouble assessing the improvement of our efforts. For example, if we did a better job of tidying a certain house, we wouldn’t know if it was due to our increased skill or the fact that the house was simpler to organize. We did the usual method of demonstrating our work and then displaying a nice video. There was no way for us to know whether our methods were really producing the desired results. We wished to set up a challenge wherein the conditions are similar to a home situation, where we can accurately evaluate how much progress has been achieved.

Although the dialogue was educational and intriguing, I am not prepared to make a 12-hour round-trip flight for a 24 hour stay (special thanks to the individuals who offer to care for my rabbit at those times). I was glad that the South Bay area was more than willing to help out. Thursday was essentially two conversations with venture capitalists from Palo Alto, each lasting an hour and a half. The morning began with Rohit Sharma from True Ventures, and I ended with Bruce Leak and Peter Barrett at Playground, a 10 minute drive away.

is a venture capital firm.

“True Ventures” is a company that invests in early-stage startups.

The photo used in this piece was provided by Current Surgical.

It was fortunate that I made contact with True Ventures because I had been writing about Bigscreen and Current Surgical, two of their companies, which had recently become successful. Neither of them operates in the robotics industry so there’s no need to spend time going into detail about them.

I’m still looking over all the tremendous wisdom I heard, but I thought I’d share some exciting remarks from the three savvy venture capitalists I chatted with.

Rohit Sharma has been referred to as a cricketing superstar.

We’re not expecting a general robotic or a humanoid robot to be wandering around providing help. That is however a decent thought to have, but it isn’t exactly a product or technology. There will always be human labor at the border areas of robotics. If, for instance, a machine is made to organize gardens and maybe even fields and carry the produce, there will still be exceptions that need to be managed by hand in every 100-ft operation. Everyone crowds around when they demonstrate their strawberry harvesting robot or their favorite harvesting robots to 10 individuals.

An area set aside for recreational activities, especially one with special equipment for children to use.

Special recognition is given to Brian Heater for the images used.

Is Bruce Leak questioning if it’s too early to create a humanoid robot that is capable of cleaning garages or shoveling snow and is accessible through an app store?

Peter Barrett stated that anything regarding Rosie the Robot is simply a 1960’s cartoon fantasy and will continue to stay that way. He continued on to say that making a successful investment should be based on knowledge of knowing where to put the money. He believes that a practical, deployable, and reliable outcome can be accomplished.

In the past four years, autonomous taxis have been moved around various places such as pavements, residential neighbourhoods, sports grounds and farms. However, BL doesn’t want any rules or people in their way, and doesn’t want them to travel at quick speeds.

P.B. stated that the difference between unloading a truck at Amazon and then getting out of the driverless car to deliver a parcel is something that can be solved through software. Although the vehicle is already equipped with the capability to perform multiple tasks, software development is lagging behind as it is not able to effectively deal with diverse external elements such as turf, hoses and animals. In other words, technology advancements in the mechatronic field are further along in comparison to the perceived lack of progress in its cognitive capabilities.

PB stated that humans will still have the ability to do many tasks that robots are not capable of completing for a long time.

Swap Robotics provided the image.

Swap Robotics, the company that completed the Finals of the Battlefield at Disrupt last year, has just revealed a $7 million seed investment. It has attracted attention for its modular systems which can do tasks such as lawn mowing and snow clearing, and specializes in solar farms. Tim Lichti, the CEO of the company, spoke to TechCrunch and said:

Solar energy is an important market for our company. We believe that using robots to trim the grass and vegetation around utility-scale installations is a great solution for customers, since this is one of the most expensive ongoing costs associated with maintaining solar sites. Since the areas around these sites are closed off to the public, electric robots can operate anytime, day or night, and help maximize the hectares cut each month. Experts anticipate the total value of solar trimming across the world reaching tens of billions each year by the 2030s.

The photograph comes from Pudu Robotics.

An additional $15 million in Series C3 funding is being given to Pudu, located in Shenzhen. This follows their 2021 Series C2 round where they raised a total of $77 million. The company is preparing for their upcoming Series C4 round, which is expected to be particularly notable. They specialize in robotic technology primarily related to the food service industry, according to the public announcement.

Felix Zhang, creator and manager of Pudu Robotics, revealed that the business has seen steady development in the delivery area with the acceleration of overseas redistributions. In the year 2022, 90% of the company’s earnings came from sales. Pudu Robotics has declined to follow quick development by launching more cost-effective versions and has stopped rented items from occupying an excessive segment of income, making sure a good performance in payments and collections. In 2022, there was a boost in income of close to 40% versus the prior year, resulting in more than 20,000 units being shipped. The high-endmodels, with the BellaBot as a prime case in point, have pulled in quite a lot of attention from international venues. Amazingly,Skylark, the biggest restaurant group in Japan, placed an order for 3,000 BellaBots simultaneously, making this the most significant transaction in the field.

Source: Microsoft

ChatGPT was a significant topic last week and continues to be so, overtaking the focus that was on blockchain robotics in the metaverse just a few weeks ago. I recently wrote about the concept of hype-cycles, and it is pertinent to consider upon reading a related story from Microsoft about utilizing ChatGPT in their robotic systems. This paper outlines the common issue of creating code to run robotic systems.

ChatGPT enables a new kind of robotics by enabling a user to give the LLM higher-level directions and observe the robot’s behaviour. This is done with no further modifications, by utilizing the LLM’s information to control a range of robots that can carry out a selection of tasks. This is all done as part of a design protocol created by ChatGPT. In our research, we demonstrate various applications of ChatGPT in resolving robotic puzzles, in addition to complex robotic deployment operations in the handling, flying, and navigation fields.

A great number of individuals are investing a significant amount of money to tackle this dilemma. Even though I support the concept of forming a system that can adeptly accomplish WYSIWYG robotic programming, right now we must see it as something with possibilities, but that may be exaggerated. However, these may be valuable paths to examine while we strive to make robots more available to people who are not roboticists.

Firms that are recruiting for robotics positions

Seven positions are offered by Righthand Robotics.

There are thirteen positions available for Chef Robotics.

Two positions are available for people to work with Impossible Metals.

Photograph by Bryce Durbin for TechCrunch

I’m on my way to Catalonia and while I’m gone, why not sign up for Actuator?