In the first prototype of the JetsonBot, there were two ‘duck’ antennas connected to an Intel 7260 Wireless/BT NIC card
You have probably seen versions of the duck antennas that were being used on the JetsonBot prototype installed on wireless routers. These types of antennas are often used on electronics which are in metal enclosures. The metal enclosure tends to shield radio waves, having external antennas overcomes this shielding issue.
Because the JetsonBot is made mostly of plastic, and is pretty much an open air device anyway, we can choose a different type of antenna. As you probably know, most laptops, tablets and phones have their antennas built in to the case. In the case of laptops, there is usually a foil antenna which as the name describes is a strip of foil that is connected to the wireless device which provides the laptop connectivity. In some cases, a device called a ‘panel antenna’ is used. For the JetsonBot, this seems like a good solution because it can be attached to the bottom of a platform and pretty much be out of the way. In the video, a Pair Laptop Wireless Mini PCI PCI-E Internal Antenna
In order to install the panel antenna, the Jetson was removed from the top platform, and the platform removed from the JetsonBot. A hole was drilled in the center of where the Jetson TK1 is mounted as a pass through to route the panel antenna wire. A counterbore was drilled in the hole (along with a little bit of sanding) to smooth the edges of the hole so that the wires would not chafe against the edges.
From underneath the platform, the two antenna wires were passed through the hole with enough slack to be able to attach to the Intel 7260 U.FL connectors. The panels on the antennas have self adhesive tape on the back to attach the antenna to the underside of the platform. After a little wire management attaching the rest of the wires to the underside of the platform, the platform is replaced on the JetsonBot and the Jetson TK1 reinstalled.
The amount of signal that the panel antennas receive compared to the duck antennas appears to be pretty much equivalent in limited testing. Both approaches seemed to be within 1 dBm of each other, neither performing better than the other on any given task. I’m also sure that better performance could be expected with a more engineered approach to antenna placement, but for now it seems to work fairly well.