Speaker
Description
Connected robotic systems perform many complex tasks through coordination, such as cooperative search of an environment, consensus, rendezvous, and formation control. At their core, these systems rely on local coordination between intelligent agents making reliable, low-latency, high-rate wireless communication of primary importance. Beyond simply maintaining connectivity, reliable communication may mean supporting heterogeneous and possibly time-varying communication rates amongst different pairs of agents. For example, some agents may need to use the network for transmitting video while others may simply wish to transmit status information. In this paper, we will describe an approach to bridge ROS and GNU Radio to enable the utilization of software-defined radios, such as Universal Software Radio Peripherals (USRPs), within the ROS ecosystem for connected robotics applications. This work is the driver of the NSF-sponsored research testbed at FAU’s Center for Connected Autonomy and AI, which focuses on networked robotics. To achieve robot-to-robot communication we will incorporate GNU Radio-based LoRa transceivers. The integration is achieved through a custom GNU Radio block that interfaces the PHY layer of gr-LoRa with the upper layers of LoRaWAN implemented in python and the data distribution service in ROS. Data is labeled based on ROS topics and transmitted using TCP ports between GNU Radio and ROS. The experimental setup involves two Clearpath Dingo indoor mobile robots, each equipped with an NVIDIA Jetson Orin and a USRP B210. We show that Dingos can share their locally generated LiDAR maps by establishing an ad-hoc wireless communication link. Link signal-to-interference plus noise ratio is optimized by dynamically controlling the relative locations of the two robots.
| Talk Length | 30 Minutes |
|---|
