Chassis and Drivetrain

Phobetor was designed to be a durable, reliable and waterproof autonomous platform that fixed many of the hardware issues plaguing the 2009 entry, Argos. The robot measures 31″ wide, 37″ long and 5′ 9.5″ tall and weighs approximately 270 lb. Phobetor can easily fit through a doorway and is about the size of the average human. Additionally, the higher position offers a superior viewpoint that allows a wide field of view to ensure better lane and obstacle detection. It has a tricycle wheelbase to ensure stability and allowing the robot to maintain contact with the ground regardless of the terrain. With a low center of gravity, the wheelbase makes Phobetor extremely stable and holonomic. Two NPC-T64 motors power the rear two wheels that are outfitted with snowblower all-terrain tires, and leading caster provides additional support in the front. The 80/20 aluminum-frame chassis of the robot is covered primarily with black MDS-Filled Nylon 6/6 and the two “wing” panels were cut out of Garolite.

Much thought was put into selecting a paneling material due to the fact that in the previous year’s entry, strong temperature fluctuations induced bowing and bending of the previous paneling material. To compensate for this, the team researched a cost-efficient alternative that could bear great temperature differences without expanding or shrinking. Two 120mm, 60cfm (cubic feet per minute) fans are installed in the rear of the robot and a weatherproof grate is positioned in the front to provide ventilation and air circulation. Like its predecessor, Phobetor has a bi-level construction to allow easy access to its electronics. The bottom layer contains the 24V electrical system, motors and Victor 885 motor controllers. The top level contains most of the electronics and features hinged “wing” panels that can be unlatched and lifted up to allow quick access to any of the devices on that level. The entirety of the exterior has been sealed using various materials in order to make the robot waterproof, allowing testing to be carried out in all but the most severe weather conditions.

Sensors and Controls

Phobetor boasts an array of absolute and differential sensors that affords it accurate state estimation. The robot, like Argos, makes use of the Gladiator G50 MEMS gyroscope which provides reliable, instantaneous yaw rate data to correct for possible wheel slippage. Two US Digital wheel encoders generate high-precision, low-drift data for angular velocity and an OceanServer OS500-US digital compass mounted in the tower provides heading data. The sole vision sensor of Phobetor is a Videre STOC (stereo-on-chip) color camera with a 15cm baseline. Both lanes and obstacles are detected on this device, avoiding the errors associated with using individual cameras for each task. The camera’s onboard Field Programmable Gate Array (FPGA) calculates the 3D point cloud, reducing computation load on the robot’s computer and allowing for a higher resolution and refresh rate. Due to the camera’s placement (about eye-level), Phobetor can detect obstacles anywhere between 20cm to 20m in front of its leading caster. Finally, the robot boasts a new wireless emergency stop system manufactured in-house and a 2.4 GHz spread-spectrum wireless radio system that allows for manual control of the robot using a traditional RC remote controller.

Software

Phobetor‘s software architecture was completely rewritten for the 2010 IGVC. The underlying robotics framework makes use of IPC++, an object-oriented wrapper of Carnegie Mellon’s Inter-Process Communication (IPC) platform. Each software component runs as a discrete programming communicating with the central serve via TCP. Message publishing and subscription, serialization and timing are all all built around a custom-developed C++ API. One of Phobetor‘s key software innovations is the use of the SRCDKF (Square Root Central Difference Kalman Filter) to combine data from all of its state sensors (such as compass, GPS, and so on) and maintain an optimal estimate of a vector that defines the state of the robot. The Videre stereo camera creates a 3D point cloud in which obstacles are detected using an algorithm that searches for points that are approximately vertical to each other. The lane detection algorithm, originally developed for the 2007 DARPA Urban Challenge, applies yellow, white and pulse width filters and, along with the obstacle detection image, fuses the resulting images together into the same frame. 

To ensure that lane markings in shadows are not ignored, a white filter operating in hue-saturation value space has been implemented. The RANSAC (Random Sample Consensus) algorithm is then used to fit parabolas to the frame, detecting lanes with great accuracy and even tolerating gaps within the image. For path planning, Phobetor uses the Anytime D* map-search algorithm to search through the cost map generated by environmental data. The algorithm allows the robot to quickly plan paths around any obstacles it may detect in real time, and if the robot senses it may have more computing time available, the software takes additional time to plan the optimum trajectory towards the desired waypoint. Additionally, the autonomous platform uses a crosstrack error navigation law to follow paths, allowing the algorithm to plan accurate, smooth movements for the robot. A speed controller is also implemented to eliminate sudden acceleration or deceleration (except in the case of emergency stops).