ASRL News

Snow Day

2016-12-15T14:47:00.003-05:00

It's been another great year for ASRL and we celebrated today with our annual group lunch. Conditions on the way back were pretty extreme, so we decided it was a good time to update our group photo. A great reminder that we have a lot of work to do before robots are able to handle all environmental conditions!

Goodbye, Mr. Anderson

2016-11-04T09:58:00.002-04:00

Sean Anderson successfully defended his PhD thesis on November 3rd, 2016; congrats to Dr. Anderson! We celebrated with a nice meal at Hart House Gallery Grill and then hung out on campus for a bit. Along the way finally got a chance to try out Rehman Merali's teaBOT loose-leaf tea robot...the tea was great and the lid is very cool since it acts as the tea strainer.

Classic Canadian Robotics

2016-08-12T12:33:00.002-04:00

Our proposal for a data collection car wasn't funded this year so we decided to throw together a rig on a personal car instead. We logged 50 km of Velodyne-64, stereo, and POS-LV data around the UofT main campus. The rig was mounted on a standard roofrack with cables coming through the sunroof and power coming in via jumper cables right on the car battery. We're resubmitting our proposal this year for a more permanent solution, but this data should get us started and the setup will inform our eventual design if/when we actually get some money.

All Night Long

2016-07-21T09:59:00.000-04:00

We just finished an experiment this week showcasing our new Multi-Experience Localization (MEL) for use in Visual Teach and Repeat (VT&R). Using only cameras, we taught a route in bright sunlight and demonstrated that we could continue to drive exactly in our tracks all night long (with the use of headlights). This works because we are localizing against many previous experiences all the time and we can therefore track the visual appearance change. The image shows the Grizzly driving in its tracks well after sunset. We're also using a drone to make videos hence the follower.

Back to Reality

2016-06-19T10:44:00.002-04:00

ASRL Field Team [photo: Pomerleau]

Our Sudbury 2016 field campaign has come to a close and we're back in civilization. All of our experiments were quite successful. The Grizzly drove over 120 km using vision-only feedback, the Lancaster planes did many surveys of the site, and TReX explored two unique rock outcrops with several descents at each one. After cleaning up all the gear, our next step is to process the 6 Tb of data we collected, post pictures, make videos, etc. Can't wait for the next field experience!

On the Rock

2016-06-11T09:19:00.001-04:00

We're having a great field campaign in Sudbury. TReX, our tethered climbing robot, has completed several descents down steep terrain in order to build three-dimensional models using lidar. The robot has performed quite well, and we've become more confident in the strength of the tether over time. The ability to drive sideways while under tension has proven an important capability to get around obstructions such as rock bulges. Overall, we're happy with the TReX concept and hope to make a smaller, lighter machine in the future to simplify logistics.

Sudbury Field Camp

2016-06-10T09:54:00.001-04:00

We're back in the Ethier gravel pit in Sudbury. We've had our camp running for 5 days now and have gotten into a good routine. The Grizzly has passed 75 km of autonomous driving using VT&R2, TReX has explored two different rock outcrops, and our fixed-wing UAV has done an aerial survey of the pit and our camp. The weather has taken a turn for the better so it's looking great for the next few days.

Trains are Robots, Too

2016-06-03T06:40:00.000-04:00

Congrats to Tyler, whose paper on lidar-based train localization won best robotics paper at the conference on Computer and Robot Vision (CRV) in Vancouver this year!

Sudbury Bound

2016-05-26T15:43:00.003-04:00

It's been two years since we were on a big field test but we're off to live in a gravel pit in Sudbury, Ontario for two weeks, once again. We'll be testing a new version of Visual Teach & Repeat (and plan to drive 100 km autonomously), a cliff-descending tethered robot for geological mapping, and some small fixed-wing planes for aerial surveillance. Image on the right shows the big pile of gear we're starting to gather for our trip. We've even got the floorplan of our office truck mocked up and have been making sure our power system will work. Hoping for no bugs, both software and of the insect variety.

Full STEAM Ahead at IROS 2015

2015-09-29T19:47:00.002-04:00

Sean Anderson presented our paper, "Full STEAM Ahead: Exactly Sparse Gaussian Process Regression for Batch Continuous-Time Trajectory Estimation on SE(3)", today at IROS 2015 in Hamburg. He got a huge audience and most of them are even looking at the screen! Time to relax and enjoy the conference.

FSR 2015 is Done

2015-07-01T08:47:00.003-04:00

Last week we hosted Field and Service Robotics (FSR) 2015 at the University of Toronto and it was a big success. The weather was beautiful for our boat cruise and live demos and all the social events were great. UTIAS presented 4 papers and they all seemed to be well received. Thanks to everybody who participated in the organization and execution of the conference; it was absolutely a team effort and the results were worth it. Just need to take care of some closing tasks and it will be done done. There's a great article on FSR at Robohub.

Grizzlies Belong in the Woods

2015-05-07T18:04:00.000-04:00

Spring has sprung and our Grizzly is on the prowl again. The image on the right shows the Grizzly autonomously repeating a route using vision at 3.5 m/s through a wooded area. This is easier said than done as there are a multitude of hazards that must be avoided. We demonstrate the route we want to drive to the robot. Then, our path-tracking controller gradually increases speed in sections of the path where it can keep errors within user-defined bounds; it therefore learns to drive quickly while keeping path-tracking errors very low.

Winter Wonderland

2015-02-03T16:25:00.001-05:00

Winter is here...brrr. We've been working on getting our visual teach and repeat navigation framework to work in some difficult conditions. Snow is one of those conditions. It's terrible for both traction (lots of wheel slip) as well as vision systems (very few stable features and changing geometry from one day to the next). Multiple cameras seems to help quite a bit and we hope to have some results of this testing submitted to the next edition of Field and Service Robotics, which will be held in Toronto in June, 2015.

T-ReX Tensions Rise

2014-11-07T20:15:00.001-05:00

T-ReX tension calibration

We're developing a new version of our climbing robot called T-ReX (Tethered Robot eXplorer). The plan is to deploy it in a canyon in a newly discovered crater on Victoria Island in the Arctic in the summer of 2015. The mission will be to teleoperate the robot and build a 3D model of the stratigraphy of the crater that has been exposed in the canyon. We've completed the mechanical design and here are calibrating the tension sensor so that we can start to implement tension-based control for ascent/descent on near-vertical surfaces. We'll be deploying the robot on the exterior of our MarsDome to emulate the canyon wall in the near future.

RSS: Ticket to Ride

2014-07-20T23:24:00.000-04:00

Last week was the Robotics: Science and Systems (RSS) conference, held in Berkeley, California. Tim presented our latest findings on viewing batch continuous-time trajectory estimation as exactly sparse Gaussian process regression and Jon presented a new sample-based path-planning algorithm, BIT*, that uses heuristics to help guide the search analogously to A* but in continuous spaces. The dinner on the last night was held at Google headquarters in Mountain View and we had a chance to go for a ride in one of the self-driving cars…very very impressive stuff!

Lab Featured on CBC's Our Toronto

2014-06-23T08:31:00.002-04:00

Our lab was featured in a news item on the CBC show, Our Toronto. The link to the episode is the
following:

UTIAS Robotics on CBC's Our Toronto

The piece occurs at about 34 minutes into the show. Thanks to Keith Whelan and Jonathan Castell from the CBC who shot and edited the story together.

Skull Crusher Debut

2014-05-17T10:44:00.003-04:00

This past week we had a team testing an improved version of stereo visual teach and repeat on the Mars Emulation Terrain in Montreal, Quebec, with our brand new Grizzly mobile robot, "Skull Crusher". We taught a 1 km path that was partly on the Mars terrain and partly in the adjacent wooded area. We were able to repeat this path over a wide variety of lighting conditions 26 times, with 23 of these being 100% autonomous and the others over 99% autonomous. Results to be submitted to ICRA 2015. Fun times!

Thriller at CRV

2014-05-10T10:38:00.003-04:00

Several members of ASRL are in Montreal this week attending the Conference on Computer and Robot Vision (CRV) and the NSERC Canadian Field Robotics Network (NCFRN) annual field trials. We had a good year at CRV with Chris Ostafew's paper winning the Best Robotics Paper Award and Chi Hay Tong's PhD thesis garnering an honourable mention for the CIPPRS Doctoral Dissertation Award. The CRV banquet was held at a wax museum, hence the pic.

IROS in Tokyo

2013-11-11T05:25:00.003-05:00

IROS in Tokyo was good this year. ASRL presented two papers and both seemed well received. The picture on the right was from our motion-compensated RANSAC talk. Glad to see some familiar faces in the crowd listening in.

On the cover of the Rollin' Stone...

2013-10-08T04:27:00.001-04:00

Every year the Journal of Field Robotics has a photo competition to select shots for its cover. This shot was from one of ASRL's field tests to the Canadian Space Agency's Mars Emulation Terrain in 2011. At the time we were testing our laser-based 3D site mapping capabilities as well as our main stereo navigation pipeline (network of paths, teach and repeat). The photo came second in the competition and will appear on the cover of all the 2014 JFR issues.

1893 Stereo Camera

2013-07-28T04:54:00.000-04:00

Tim is on sabbatical in the Mobile Robotics Group at Oxford until January. There is a wealth of interesting history in Oxford. The image on the right shows an early stereo camera on display in the History of Science museum. Notice the gear in the middle used to simultaneously adjust the aperture on both cameras.

CREATE Deployment at CSA Mars Emulation Terrain

2013-06-06T16:55:00.002-04:00

This week ASRL members participated in a mock Mars sample return mission at the Canadian Space Agency in Montreal. Our role was to support the science team in planning and executing rover traverses. We also demonstrated our new vertical-surface robotic explorer, vScout, on some of the rubble and cliff faces on Mars Emulation Terrain. In all it was a success week and was a great collaboration between academia, government, and industry.

ASRL AR Drone Video Mentioned on IEEE Spectrum

2013-06-04T11:07:00.001-04:00

Here's the link: ASRL AR Drone Video on IEEE Spectrum

Chi Passes FOE

2013-04-30T09:24:00.002-04:00

Yesterday, Chi Hay Tong passed his PhD final oral examination (FOE). Congratulations, Dr. Tong! Prof. Frank Dellaert, world-renowned expert on simultaneous localization and mapping from Georgia Tech, was the external examiner; he asked lots of excellent questions! Prof. Radford Neal from UofT CS was the internal examiner and also weighed in with some fantastic questions about Gaussian Processes. In the end, Chi emerged with a PhD and was rewarded with this very nice cookie, courtesy of the UofT Institute for Robotics and Mechatronics. Chi will be hanging around ASRL for the next couple of months and then heading off into the world!

vSCOUT: First Ascent

2013-04-27T21:25:00.002-04:00

We've been working with geologists from Western for several years now to look at how geological investigations could be done on other planets using robotics. One thing we've learned is that a lot of the interesting places are not on flat ground but rather up on vertical surfaces where stratigraphy, for example, can be seen. Enter vSCOUT, a mobile robot with a powered rope ascender to allow it to drive on near-vertical surfaces. Today we demonstrated that we could successfully drive up and down the MarsDome at UTIAS under manual control. The next step will be to do this autonomously using visual navigation and eventually we want to build photorealistic 3D models and more.