Last day of IV2010

My last post from IV2010 sort of got out of sync because of all the traveling. In the morning the single-track session covered mainly perception but also V2V communication. In the afternoon keynote speech Dr Uwe Franke gave quite an interesting talk about the research evolution in stereo vision and concluded his talk by underlining the need for integrating communication and perception technologies for effective future active safety systems.

The day closed with an outdoor demo of a number of research vehicles and it was great to get to see these up-close. I especially did a private survey of trunk installations as we need to start thinking about such things for our upcoming GCDC involvement. Looks like we'll have to invest in some tie-downs ;)

Stanford/VW autonomous valet parking vehicle

A UCSD data collection vehicle

Another VW if I remember correctly.

IV2010 Day 1

Today the 2010 Intelligent Vehicles Symposium kicked off with four special sessions and a PhD dissertation forum. Of the two special sessions in the morning one was a tutorial on automotive software architectures and the other was on wireless communications and intelligent vehicles. I attended the latter one and there were quite a few interesting talks.

More than one presenter focused on connectivity in multihop vehicular networks and how to adjust parameters such as transmission range to achieve a good trade-off between connectivity and contention.

There was also an interesting talk by Justin McNew from Kapsch comparing 802.11p direct V2V communication with infrastructure (LTE) based communication. Capacity calculations for the two alternatives indicate that capacity will be an issue in the LTE setting due to the large number of nodes that need to be serviced by each base-station.

The session ended with a panel discussion featuring Richard Bishop, Alexander Busch and Justin McNew. Among other things Richard Bishop brought up the situation that regulatory organizations find themselves in regarding mandating V2V communications equipment. For "traditional" safety systems such as airbags and ESC it is easier to push mandates as the advantages of such systems can be clearly shown through testing, however the benefits of mandating cooperative safety systems are not as easily shown.


The afternoon session on human-centered intelligent vehicles focused partly on the integration of drivers with partially autonomous systems and how to share responsibility between the two. Will we see systems where the driver is tasked with high-level planning (e.g. "I want to change to the left lane") and the vehicles takes care of implementing those driver decisions? How can such systems be created without losing the advantage of having a single universal vehicle control interface?

The day ended with a PhD dissertation forum with short, 10-minute, presentations with feedback from seniors in the area. With 14 presenters back-to-back this was quite intense but very interesting, as I move closer to dissertation (hopefully ;) ) next year, this would be a good venue for feedback.