Planning

Introduction to the Planning Module

The autonomous driving pipeline encompasses a multitude of components, from perception to control. Central to this pipeline is the “Planning module”, responsible for making strategic and tactical decisions about the vehicle’s movement. This module breaks down into three key stages: Route Planning, Behavior Planning, and Motion Planning, planning the movement of the vehicle at different levels of granularity and accuracy. In this post, we will explore each stage in-depth.

kine_bicycle_model

Pipeline of planning and control

Route Planning

Objective

The task of route planning stage is to find the most effective course from the current location to the desired destination, much like how conventional GPS navigation systems operate. Modern route planning algorithms are able to return an optimal route on large scale in real time, after a one-time pre-processing step [1]. For a more comprehensive survey, refer to the paper by Bast et al [2].

Inputs

The route planning stage takes as input high-level information about the world, such as:

  • the coordinates of the start point and of the end point;
  • the topology of the road network;
  • the traffic conditions; and
  • possible permanent obstacles.

Outputs

The route planning stage generates as output a path or a series of waypoints bridging the start and destination, while sidestepping any permanent obstacles.

Challenges

There are many challenges the road planning module has to take into consideration when computing the most effective course, such as

  • adapting to changing road situations;
  • incorporating real-time traffic details; and
  • ensuring the chosen route is both safe and efficient.

Behavior Planning

Objective

The main task of the behavior planning stage is to dictate the vehicle’s overarching behavior, from decisions about lane switches, overtaking, yielding, to making turns. Nowadays, the behavior planning modules often incorporate the interaction between the ego vehicle and other agents on the road, such as other vehicles [3].

Inputs

To achieve its goal, the behavior planning stage uses information from multiple sources, such as

  • the designated route obtained from the route planning;
  • data about neighboring vehicles and pedestrians (sourced from LiDAR, radar, cameras, etc.);
  • traffic light information; and
  • other pertinent environmental data.

Outputs

The output of the behavior planning consists of macro-level directives like “shift to the left lane”, “trail the vehicle in front”, or “get ready to turn right” as well as coarse trajectories that reflect these directives.

Challenges

In making its decisions, the behavior planning stage needs to address multiple issues, such as

  • anticipating the intentions and forthcoming actions of surrounding drivers and pedestrians;
  • adopting a safe decision formulation in intricate scenarios; and
  • navigating uncommon edge cases.

Motion Planning

Objective

The last stage of the planning module workflow is motion planning, whose aim is to chart a trajectory that is both safe and consistent for the vehicle, reflecting the behavior objectives and staying within the vehicle’s dynamic and kinematic constraints [1].

Inputs

To produce a smooth trajectory for the vehicle, the motion planning needs multiple types of data, like

  • the intended behavior established by the behavior planning;
  • the present vehicle state (such as position, speed, and acceleration);
  • the road map details; and
  • data about dynamic obstructions.

Outputs

The outcome of the motion planning stage is an optimized trajectory, detailing the vehicle’s state over a specified time frame.

Challenges

To produce such an optimized trajectory, the motion planning faces multiple challenges, such as

  • crafting trajectories that avoid collisions and are streamlined;
  • managing noisy and uncertain data from sensors; and
  • real-time trajectory mapping.

References

  1. Paden, Brian, et al. "A survey of motion planning and control techniques for self-driving urban vehicles." IEEE Transactions on intelligent vehicles 1.1 (2016): 33-55.
  2. Bast, Hannah, et al. "Route planning in transportation networks." Algorithm engineering: Selected results and surveys (2016): 19-80.
  3. Ding, Wenchao, et al. "Epsilon: An efficient planning system for automated vehicles in highly interactive environments." IEEE Transactions on Robotics 38.2 (2021): 1118-1138.

Table of contents