Let’s try this again: SpaceX is looking to launch its 18th International Space Station (ISS) resupply mission on Thursday at 6:01 PM ET (3:01 PM PT), a day after it tried to do so a first time. The initial attempt was scrubbed at the last minute due to weather conditions on the launch range in Florida at Cape Canaveral Air Force Station, but there’s a backup window today so Elon’s rocket company will try again. The stream above will begin about 15 minutes beforehand.

CRS-18 will send a Dragon cargo capsule with 5,000 lbs of experiment materials, supplies and other cargo including a new automated docking sleeve to the ISS. This mission is one of the last in SpaceX’s current ISS resupply contract, which covers 20 total engagements – but NASA and SpaceX essentially re-upped for a second batch through 2024 as part of a new contract signed in 2016.

The Dragon cargo craft used on this mission has actually flown two previous ISS resupply missions, and will be the first to manage a third should this trip go as planned. It’ll rendezvous with the ISS in a few days, be unloaded by astronauts and re-loaded with return cargo, before detaching from the ISS and returning to Earth to splash down in the Pacific Ocean for recovery. Dragon will launch aboard a Falcon 9 first-stage booster that was used just a couple of months ago for the most recent CRS mission, too.

Alphabet’s autonomous driving and robotaxi company Waymo does a lot of training in order to refine and improve the artificial intelligence that powers its self-driving software. Recently, it teamed up with fellow Alphabet company and AI specialist DeepMind to develop new training methods that would help make its training better and more efficient.

The two worked together to bring a training method called ‘Population Based Training’ (PBT for short) to bear on Waymo’s challenge of building better virtual drivers, and the results were impressive – DeepMind says in a blog post that using PBT decreased false positives in a network that identifies and places boxes around pedestrians, bicyclists and motorcyclists spotted by a Waymo vehicle’s many sensors by 24 percent. Not only that, but is also resulted in savings in terms of both training time and resources, using about 50% of both compared to standard methods that Waymo was using previously.

To step back a little, let’s look at what PBT even is. Basically, it’s a method of training that takes its cues from how Darwinian evolution works. Neural nets essentially work by trying something and then measuring those results against some kind of standard to see if their attempt is more ‘right’ or more ‘wrong’ based on the desired outcome. In the training methods that Waymo was using, they’d have multiple neural nets working independently on the same task, all with varied degrees of what’s known as a “learning rate,” or the degree to which they can deviate in their approach each time they attempt a task (like identifying objects in a image, for instance). A higher learning rate means much more variety in terms of the quality of the outcome, but that swings both ways – a lower learning rate means much steadier progress, but a low likelihood of getting big positive jumps in performance.

But all that comparative training requires a huge amount of resources, and sorting the good from the bad in terms of which are working out relies on either the gut feeling of individual engineers, or massive-scale search with a manual component involved where engineers ‘weed out’ the worse performing neural nets to free up processing capabilities for better ones.

What DeepMind and Waymo did with this experiment was essentially automated that weeding, automatically killing the ‘bad’ training and replacing them with better-performing spin-offs of the best-in-class networks running the task. That’s where evolution comes in, since it’s kind of a process of artificial natural selection. Yes, that does make sense – read it again.

In order to avoid potential pitfalls with this method, DeepMind tweaked some aspects after early research, including evaluating models on fast, 15-minute intervals, building out strong validation criteria and example sets to ensure that tests really were building better-performing neural nets for the real world, and not just good pattern-recognition engines for the specific data they’d been fed.

Finally, the companies also developed a sort of ‘island population’ approach by building sub-populations of neural nets that only competed with one another in limited groups, similar to how animal populations cut off from larger groups (ie, limited to islands) develop far-different and sometimes better-adapted characteristics vs. their large land mass cousins.

Overall, it’s a super interesting look at how deep learning and artificial intelligence can have a real impact on technology that already is, in some cases, and will soon be even much more involved in our daily lives.

With as successful launch from the Gobi desert blasting off at around 1:10 PM Beijing time (1:10 AM ET), Chinese space launch startup iSpace (which, awesomely, is also called ‘StarCraft Glory Space Technology Co.’) became the first private Chinese commercial space launch provider. The company’s SQX-1 Y1 rocket delivered two commercial satellites to an orbit of about 300 km (about 186 miles) above Earth.

The launch is the first successful commercial mission for the SQX-1 Y1 solid-propellant rocket developed by iSpace, which is a four-stage design that can carry up to 260 kg (around 575 lbs) and weights around 68,000 lbs.

This is a major milestone for the Chinese space industry, and iSpace beats out a healthy crop of competitors, including LandSpace and OneSpace, both of which did not succeed in earlier attempt to be the first in China to the private launch market.

iSpace, founded in October 2016, has secured a Series A funding round of an undisclosed amount in June, including invent from CDH Investment, Matrix Partners China and Shunwei Capital . The company previously completed sub-orbital flights of precursors to the SQX-1 Y1 rocket in 2018.

SpaceX encountered a snag in an attempted test key to the development of its next-generation Starship spacecraft. Specifically, the StarHopper sub scale demonstration and testing craft it’s using to work on the Starthip’s propulsion system fails to undertake its first untethered test flight at a testing site in Boca Chica Beach in Texas,

The plan was to have the demonstration craft take off and fly to a height of 20 meters before returning to Earth, all under tis own power and directed by its own guidance system. Instead, It seemed to fire rockets and then was engulfed in smoke, before venting fire out of the top of the test craft for a few minutes prior to extinguishing, with StarHopper looking relatively unscathed. We’re still waiting on official confirmation of what happened from SpaceX, but they characterized this as an “abort” on a livestream of the test.

Last week during a static test fire, the StarHopper vehicle was engulfed in a large ball of flame. This wasn’t a planned event, but did not result in significant damage to the spacecraft, SpaceX later said.

StarHopper succeeded in flying its first tethered flight at the beginning of April, and has undergone further testing since then to prepare for its untethered trip. SpaceX CEO Elon Musk said earlier this month that a successful untethered test would pave the way for a full presentation of SpaceX’s Starship spacecraft plans at the end of July, but the test has encountered a few issues since then.

The reason SpaceX and other companies run tests like these is to identify potential issues early in the development process, so it’s good to see them making progress even if that doesn’t mean a “success” in the traditional sense of actually having achieved untethered flight.

SpaceX designed Starship will be fully reusable once complete, unlike Falcon 9 and Falcon Heavy, so it’ll reduce the cost of launches, and the company hopes to eventually use it to fly all its missions, though it’ll keep Falcon 9 and Falcon Heavy in service for its paying customers as long as there’s appetite.

Tesla is set to ramp up the rate at which it opens new service facilities aggressively, according to CEO Elon Musk’s guidance on the company’s Q2 2019 earnings call. In total, Tesla opened 25 new service centers during the quarter, and added 100 new service vehicles to its existing fleet – which is in contrast to an earlier statement made by Musk that they’d look to close most of their physical stores in an effort to reduce costs.

Notably, Musk referred to the locations only as ‘service centers’ during his comments on the subject on Wednesday’s earnings call, and never as stores – asked about ‘retail locations,’ he corrected the analyst asking and again said that what Tesla opened were ‘service centers’ specifically. He also emphasized the importance of ensuring that service scales in line with the size of Tesla’s overall fleet of vehicles in active use. Musk mentioned that the number of Tesla cars on the road doubled in the last year alone, meaning its seeing exponential growth in terms of the total size of the fleet it needs to service.

“Service scales not just with new production, but as the whole fleet sales,” Musk said, adding that they want to grow their service capabilities in a way that’s responsible when it comes to cost, but that that is “quite difficult” when it comes to the rate at which the company’s sales and shipments are increasing.

Even so, Tesla is taking on still more of its service work itself, rather than outsourcing to external vendors.

“We’ve in-sourced a great deal of the collision repair activities, which I think had quite a good impact on customer happiness,” Musk said. “This will continue in the months to come.” Musk also noted that the company is working hard to reset its processes in order to ensure that parts are available on-hand when and where needed for service, which is a gap that has prompted customer complaints in the past.

The Tesla CEO said that he meets with the Tesla service team “multiple times a week” to “get updates on the reliability of the vehicle,” noting the the best service possible is “no service” because that would represent maximum reliability (and of course, lowest possible ongoing costs for Tesla). He also said that the’ve seen “fewer and fewer service visits” “for the most recent cars that we’re building, so we’re on a good trend there.”

Tesla Head of Investor Relations Martin Viecha also noted that the number one reason for service visits is actually people looking to learn how to use Autopilot, and in general education represents a high percentage of visits.

Tesla CFO Zach Kirkhorn addressed a question about the service center expansion later in the call adding that the company is pursuing a path of systematic “focus on service and supercharging, as opposed to a retail presence.” He also noted that he believes efforts to improve their parts distribution, with a focus on ensuring that parts are available on-hand in inventory at the service centers where they’re needed will actually help bring down costs overall vs. housing them centrally or ordering on-demand from suppliers and Tesla’s own fabrication facilities.

Tesla has already started the preparations required to get production started on its forthcoming Model Y compact all-electric SUV, according to Tesla CEO Elon Musk . Musk noted that prep had started at its facility in Fremont during his introductory comments on the automaker’s Q2 2019 earnings call, confirming a report from CNBC from March.

In Tesla’s first earnings call for 2019, Musk said that it was in the process of deciding between Fremont and its Gigafactory in Nevada for production of the Model Y, which is going to be based on the Model 3 platform and will share some of its componentry, something that Musk noted will help reduce its cost of production.

The Model Y, revealed in March, looks quite similar at first glance to the Model 3. It has a slightly higher profile, however, putting it in this compact SUV range. It has similar interior features to the Model 3, including the horizontal 15-inch touchscreen, and also features a panoramic roof more like its larger Model X premium all-electric SUV sibling. Pricing for the Model Y will begin at $39,000, and that version will have a 230-mile range. It’s currently planned to ship sometime in the fall of 2020.

In today’s instalment of ‘the future is 100% subscription-based,’ Toronto-based startup Rover is testing out subscriptions for its parking marketplace. Rover lets users list their unused parking spots for on-demand rental by others on the service, giving them a passive way to earn some income while hopefully increasing the utilization rate of parking spaces at the same time.

Rover has offered the spots on their platform on a per use, on-demand basis before now, but it’s going to pilot a monthly subscription starting this summer, with a planned test phase extending into early fall. The company says it’s going to try out a few different versions of a monthly sub, including potential perks like a percentage discount vs. individual on-demand parking charges, advanced booking and premium customer service.

Pricing should be in the ballpark of between $5 and $15 Canadian depending on the features you’re willing to pay for, and this should inform eventual subscription price points for the startup’s services should they move beyond this pilot phase. Rover currently offers spots in Toronto, Montreal and Ottawa, with plans to expand to Canada’s west coast and then eventually California in future.

Uber recently debuted a subscription pilot that rolls in its ride-hailing, Eats, bikes and scooter rental services, and Rover cites this move as an example of the move to subscriptions generally in the on-demand space in its own announcement. Subscriptions are a great way for consumers to easily take car of known recurring costs, but the rise of this business model across a range of industries will definitely test the limits of consumer willingness to trade cost for convenience.

SpaceX is currently set to launch its eighteenth commercial resupply mission (CRS-18) for the International Space Station at 6:24 PM ET (3:24 PM PT) on Wednesday, from LC-40 at Cape Canaveral Air Force Station in Florida. The mission will carry equipment for experiments and scientific research, as well as supplies for ISS astronauts, and a new docking adapter that will automate the docking process for any future crew spacecraft built to take advantage during their visits to the ISS.

Live coverage will begin via the stream above at around 15 minutes prior to the target launch time, provided all goes to plan. The standard pre-launch weather assessment isn’t looking especially good however – the USAF 5th Space Wing, which is responsible for calling the weather conditions, said that there was a 30 percent chance of favorable weather as of earlier this week. Should the launch be scrubbed for today, there’s a backup window on Thursday, July 25 at 6:01 PM ET (3:01 PM PT).

SpaceX’s Dragon capsule used for this mission is badged with an “Apollo 50th” commemorative graphic near the side-hatch on the spacecraft, to celebrate the 50th anniversary of the Apollo 11 Moon landing mission. That anniversary actually ends today as well, with the returned astronauts splashing down in the Pacific Ocean in the Columbia command capsule on July 24, 1969.

While the delivery of the new automated docking module is a big highlight for this mission from the perspective of the overall space program, there’s another big milestone for SpaceX – the Falcon 9 first stage used in this mission will attempt to land back at LZ-1 at Cape Canaveral, and it’s already flown and been recovered twice previously.

UPS has big plans for drone delivery, and it’s taking two key steps to put them into action. First, it’s building its own dedicated subsidiary focused entirely on drone delivery called UPS Flight Forward, and it’s seeking FAA approval to operate its drones over populated areas, during nighttime hours and when not within view of a human operator, all of which are currently required for general commercial drone operation.

UPS will seek to gain the same certification that Alphabet’s Wing received back in April, which is a status that others including Uber Eats and Amazon Air have applied for but not yet received, as noted by The Verge.

One of the biggest shipping companies in the world, UPS basically needs to have some kind of play in the drone delivery field, whether or not that actually ends up being the future of last-mile logistics. Amazon, as mentioned, is seeking similar approval, and has been very aggressive in terms of their marketing and promotion of their drone delivery efforts.

Earlier this year, UPS partnered with drone startup Matternet to pilot medical sample deliveries in North Carolina, and the company demonstrated drones delivering packages from trucks back in 2017 in Florida, although the tests actually didn’t go as smoothly as UPS probably would’ve liked.

It’s nectar exactly how long it will take UPS to get approval, but the carrier seems confident it’ll be before year’s end, at which time we might see more in terms of actual commercial delivery service by drone rolling out.

Crowdfunded spacecraft LightSail 2 is making good on its name, after successfully unfurling its solar sail in orbit so that it can begin propelling itself using the force of light alone. The sail’s mylar surface reflects photons from the Sun, accumulating velocity gradually thanks to the additive effect of countless sub-atomic impacts. The team confirmed sail deployment initiated at 11:47 AM PT (2:47 PM ET), and full sail deployment complete at 11:50 AM PT (2:50 PM ET).

LightSail 2 got its ride to space with the Falcon Heavy launch on June 25, sharing a ride with a variety of payloads including NASA and Air Force experiments. The spacecraft is the product of The Planetary Society, a non-profit organization devoted to the advancement of space exploration that’s led by Bill Nye . Its goal is to study solar sailing in practice – a technology whose conception goes back centuries, but whose actual field use is extremely limited, with only a few examples existing previously, including JAXA’s IKAROS mission from 2010.

The sail’s total propulsion power is astonishingly small, despite its size (it’s about the size of a boxing ring) – it provides about as much power as a housefuls landing on your hand. But it also will never theoretically run out of fuel, and can gradually increase its speed over time to very high velocities thanks to the friction-free environment of the vacuum of space.

“Things are great, things are nominal,” explained The Planetary Society’s Chief Scientist Bruce Betts on a live stream of the sail’s deployment, indicating that everything is to plan so far. The Planetary Society will attempt to get back images from the deployment the next time the craft is within range of a grand station, and we’ll update when those become available.

Transitioning between different types of professional roles can be challenging — new expectations arise, and fresh ways of seeing the world are required to be successful. A few weeks ago, I interviewed Anjul Bhambhri of Adobe about how she transitioned from engineering into product, where she now leads the company’s customer experience cloud.

This week, I chatted with Toby Russell, the co-CEO of online used car marketplace Shift, which generated $135 million in revenue last year and has raised $293 million in venture capital according to Crunchbase. The company is targeting an IPO in 2021.

While we have written a lot about Shift, we have written far less about Russell, whose career spans a model he calls “learn, earn, and serve.” He received a PhD in international relations at Oxford, switched to founding a startup, then served in the Obama Energy Department, before heading over to Capital One to lead digital and eventually rejoining his friend George Arison and Shift’s co-founder as co-CEO.

Across these roles, Russell has had to adapt to very different environments, and we chatted about those transitions, his lessons learned, and his approach to leadership.

Since the launch of its first electric scooter in 2015, Gogoro co-founder and CEO Horace Luke has frequently been asked when the startup is going to expand beyond Taiwan. In its home country, Gogoro’s two-wheel vehicles, with their distinctive swappable battery system, are now the top-selling electric scooters.

But Luke says the company has always seen itself as a platform company, with the ultimate goal of providing a turnkey solution for energy-efficient vehicles. Now with the launch of GoShare*, its new vehicle-sharing platform, and partnerships with manufacturers such as Yamaha, Gogoro is ready to go global.

Founded by Luke, HTC’s former chief innovation officer, and chief technology officer Matt Taylor in 2011, Gogoro develops most of its technology in-house, including scooter motors, telematics units, backend servers and software. GoShare’s pilot program will launch next month in Taoyuan City, where Gogoro’s research and development center is located, with the goal of expanding with partners into cities around the world over the next year, starting in Europe, Australia and Southeast Asia.

“Gogoro has always been out with a thesis that we will be a platform enabler,” Luke told Extra Crunch during an interview in the company’s Taipei City headquarters. “Now you’ve seen the transformation of the company. Doing something this big, like what Gogoro is doing, takes time.”

Since the release of Gogoro’s first Smartscooter in 2015, the company says it has become the best-selling brand of electric two-wheel vehicles in Taiwan, holding a 17 percent share of the country’s vehicle market, including gas vehicles.

Last year, the company began licensing its technology to manufacturers Yamaha, Aeon and PGO to produce scooters that run on Gogoro’s batteries and charging infrastructure. It also has a partnership with Coup, the European electric-scooter sharing startup that plans to increase its fleet to more than 5,000 scooters on the streets of Paris, Berlin, Madrid and Tübingen this year, and is seeking similar deals with other vehicle-sharing services, as well as local governments that want to reduce traffic and pollution (the GoShare pilot program is being launched in collaboration with Taoyuan City’s government).

GoShare’s platform is meant to be a “very robust and cost-effective, very worry-free solution for municipalities and entrepreneurs,” Luke says. Parts of the system can be licensed separately or packaged as a turnkey solution that can be deployed in as little as two weeks.

The company describes GoShare as a “mobility solution.” When asked if this means the platform can be used for other electric vehicles, including cars, Luke says “just think of us as batteries and a motor.”

“It’s just like computers and processing ram,” he adds. “It can be any form factor. It just happens to be that the two-wheel form factor is the one we’re working on and focusing on at the moment.”