29th February was a special day for us at Geoawesomeness, as we hosted our first Twitter Q&A #GeoChat with John Ristevski, Vice President of Reality Capture and Processing at HERE (Related: Calling all map lovers: What’s a #geochat and why you should participate). We had an awesome time during #GeoChat and in case you missed it, here’s a chronological summary of what happened.
The never ending waiting to kick-off the chat
Another 10 mins to go for #geochat and we are like https://t.co/uChVtQ3iGx pic.twitter.com/0UBCDg9HTy
— Geoawesomeness (@geoawesomeness) February 29, 2016
Good evening everyone! Welcome to #geochat. We are all set 🙂 https://t.co/WLdxgKpkzH
— Geoawesomeness (@geoawesomeness) February 29, 2016
Question 1:
Q1 #geochat @jaristevski When we say maps for machines, what exactly do we mean “machines” here & how different are they from maps we use?
— Geoawesomeness (@geoawesomeness) February 29, 2016
A1: could be an autonomous car or a drone or a robot #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
A1: machine maps must be more accurate and less abstract. Machines can interpret more information #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
The Twitter community is alive 😉
.@tillnm that's a great question! We can filter out moving objects after just one pass #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
https://twitter.com/cartocalypse/status/704585875733225472
.@abuczkowski maps see beyond the sensing horizon of the machine #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
Question 2:
Q2 #geochat @jaristevski How we do we create these HD maps? What level of accuracy and completeness is required by an #autonomouscar
— Geoawesomeness (@geoawesomeness) February 29, 2016
A2: we seed the creation of the HD Map with data from our True vehicle fleet equipped with LIDAR & cameras https://t.co/iyC05Pe1F4 #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
@geoawesomeness radar and sonar are also useful for use cases like collision detection but not so much for mapmaking or guidance #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
@geoawesomeness that's unclear. It will evolve as the technology and regulatory environment matures https://t.co/OuEGWdygvg #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
Machines can use attributes like road slope and curvature. Humans just need to know how to get there #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
20 mins have already go by! Twitter Q&As can be really fast 🙂
Question 3:
Q3 #geochat @jaristevski How we do prepare #autonomouscars to handle the unexpected? Can map makers be held accountable for accidents?
— Geoawesomeness (@geoawesomeness) February 29, 2016
A3: the more real-time the map is, the better prepared the machine will be for what it encounters #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
A3: the key to keeping the map up-to-date is harnessing the sensor stream from connected vehicles with rich sensors #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
Aleks decided to go rogue 😉 V2V Communications, Open Data and more…
The cars of the future will be able talk to each other #v2v. How will it influence future map making? #Geochat @jaristevski
— Aleks Buczkowski (@abuczkowski) February 29, 2016
V2V plays a big role for time-critical events but unlikely for mapmaking where V2-cloud will prevail due to aggregation #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
How are AV w/ different datasets i.e #opensource and different mapping platforms going 2 talk 2 each other? @jaristevski #GeoChat @Eddysoft
— Nettie Ginocchetti ????? (@nettiequette) February 29, 2016
Check out our sensor ingestion protocol as an example of how we streamline communication https://t.co/qAuAXbP3HA #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
Its time for Question 4:
Q4 #geochat @jaristevski How do we go about teaching #autonomouscars to drive along others & in diverse driving cultures eg. India & Germany
— Geoawesomeness (@geoawesomeness) February 29, 2016
A4: Autonomous cars need to understand human behavior to share the road. Big data from connected cars helps to reveal the patterns #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
A4: I was just in India and there were no lane markings and the rules of the road are "flexible" #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
A4: the map is even more important in places like India where sensor visibility is limited and the map can provide context #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
A little fun on the side
@nettiequette Wonder if #autonomouscars will have “cultural” and language settings for comms with humans 😉 haha #GeoChat
— Geoawesomeness (@geoawesomeness) February 29, 2016
Back to the serious stuff…
.@jaristevski it means that it is a continuos learning process, maybe even sort of #AI. Can you also teach machines to make maps? #GeoChat
— Aleks Buczkowski (@abuczkowski) February 29, 2016
Can AI power the maps? Question 5:
.@abuczkowski has nailed Q5 #geochat @jaristevski Can we use #AI to map the world in HD? https://t.co/ej10Ek1teJ
— Geoawesomeness (@geoawesomeness) February 29, 2016
Yes! We use #machinelearning to mine the data collected from our mapping cars to extract HD map features #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
.@jaristevski I guess that today it is still a semi-manual process. Will it ever be fully automated? #geochat
— Aleks Buczkowski (@abuczkowski) February 29, 2016
It will never be perfectly automated. People will always be part of the process #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
Okay, its close to the end… Lets make things a bit more relaxed. Speaking of which,
Q6 #geochat @jaristevski If robots take over the world, can the Resistance upload false maps to save the world 😉 pic.twitter.com/pqr6psYTkF
— Geoawesomeness (@geoawesomeness) February 29, 2016
@jaristevski Okay, so the resistance has to time-travel then! Damn, thought we had a movie idea here 😉 #GeoChat
— Geoawesomeness (@geoawesomeness) February 29, 2016
Time to look into the crystal ball
When driverless cars will hit the roads all over the world in a massive scale? What is still needed/missing? #geochat @jaristevski
— Aleks Buczkowski (@abuczkowski) February 29, 2016
My 2 cents: 5 years for the first truly autonomous cars, 10 years in a meaningful scale #GeoChat
— John Ristevski (@jaristevski) February 29, 2016
And curtains!!
A big hurray for @jaristevski @here & everyone who participated in the 1st #GeoChat. Have a nice evening everyone 🙂 https://t.co/dcifzykRvo
— Geoawesomeness (@geoawesomeness) February 29, 2016
Stay tuned for the next #GeoChat 😉
#Featured
Next article
The number of shark attacks in The New South Wales state in Australia has been increasing for a number of years. In 2015, 13 attacks have been recorded – more than double the number for 2014 and 2013 combined and much more than ever before. It forced The New South Wales government to search for new ways to protect beachgoers.
The rescue is supposed to come from the sky. The state is testing a shark-spotting drone that is using advanced vision capabilities to detect potential risk. In addition, the trial is looking at using the drones to deliver Life Saving Pods to people in emergency situations. The $250.000 drone developed for that purpose is the long-range, battery-powered, military grade device that is capable of up to one hour of flight time per charge.
The New South Wales premier Mike Baird described the trial as an “innovative” step in keeping the beach safe. “This technology has the potential to improve the way our emergency services respond when people find themselves in trouble,” he said in the statement.
