Docking of Autonomous Vehicles: A Comparison of Model-Independent Guidance Methods

G. Nejat and B. Benhabib

University of Toronto, Canada

Fundament Docking Problem

The Fundamental Docking Problem is to match the pose or the pose and velocity of a dock, which is accomplished in two sub-phases:

Short-Range Positioning
Final (Physical Docking)

The focus of this paper is on Short-Range Positioning, which can be achieved via:

Sensing:

Joint-Space
Task-Space

and

Path Planning:

Visual Servoing
Guidance Algorithms (Heuristic, Ad-Hoc)

Multi-LOS Task-Space Sensing

LOS sensors

LOS table

Sensory Information for Docking

Use of only Internal Sensors (e.g., odometers) for vehicle localization results in the accumulation of systematic errors over distance traveled, hence, erroneous docking.

Use of External Sensors (e.g., range finders) for direct non-contact sensing of the vehicle can reduce the impact of systematic errors. Therefore, resulting in accurate docking

Task-Space Sensing

Majority of Task-Space Sening Systems:

measure position with high accuracies,
BUT cannot measure orientation with similar precision.

Hence, an alternative LOS-based Guidance Methodology is proposed herein:

offsets are measured in the individual sensors frames of references, and
then transformed into required vehicle movements in world coordinates to achieve desired pose of vehicle within required accuracies.

PROPOSED GUIDANCE METHODS

Step 1

PROPOSED GUIDANCE METHODS

LOS-BASED GUIDANCE

Objective of LOS sensing

MODEL-INDEPENDENT PASSIVE SENSING BASED GUIDANCE

Passive sensing method

STEP 1: MEASURE PSD OFFSETS

Step 1 of passive method

GRADIENT METHOD

Objective to Guide Vehicle to its desired pose

Sub-Step 1: Determine Rotational Motion
Sub-Step 2: Determine Translational Motion
Motion Commands: Move Vehicle by Determined Motion Commands
Continuing until Vehicle has reached its Desired Pose within Tolerances.

SUB-STEP 1:ROTATIONAL MOTION

SUB-STEP 2: TRANSLATIONAL MOTION

Step 2 of gradient

Motion Commands

MODEL-INDEPENDENT ACTIVE SENSING BASED GUIDANCE

STAGE 1: LOCK LOS

active sensing stage 1

STAGE 2: DETERMINE GUIDANCE TRAJECTORY

stage 2 active sensing

STAGE 3: VEHICLE IS GUIDED ALONG TRAJECTORY

stage 3 of active sensing

SIMULATIONS (6 DOF)

SIMULATION RESULTS FOR PASSIVE SENSING GUIDANCE METHOD

simulation results passive sensing

SIMULATION RESULTS FOR ACTIVE SENSING GUIDANCE METHOD

simulation results for active sensing

CONCLUSIONS