24. State Signal Specification

State Signal Specification

Revision 0.27.0

1. Overview

The Wheelly state is represented by three signals:

Signal	Type	Shape	Values	Description
sensor	FLOAT	[1]	-135 ... +135	Head direction relative to the map
canMoveStates	INTEGER	[1]	0 ... 5	Movement and contact state
map	INTEGER	[4+N, W, H]	0, 1	Map and marker information

For a batch of n states:

Signal	Shape
sensor	[n, 1]
canMoveStates	[n, 1]
map	[n, 4+N, W, H]

Where:

• N = number of configured markers
• W = map width
• H = map height

---

2. sensor

The sensor signal represents the robot head direction relative to the map direction.

Value	Meaning
-135°	135° in the negative direction
0°	Aligned with the map
+135°	135° in the positive direction

The value is expressed in degrees.

---

3. canMoveStates

The canMoveStates signal represents the robot movement and contact condition.

Code	Description
0	Blocked with front contact
1	Front obstacle with contact
2	Rear contact
3	No contact
4	Blocked without front contact
5	Front obstacle without contact

The state depends on:

• whether the robot can move forward;
• whether the robot can move backward;
• whether the front sensor detects an obstacle.

Decision Table

Forward	Backward	Front sensor	State
Yes	Yes	—	3
Yes	No	—	2
No	Yes	Yes	5
No	Yes	No	1
No	No	Yes	4
No	No	No	0

---

4. map

The map signal is a binary multi-channel grid.

Each cell contains either:

• 0 — condition not present
• 1 — condition present

The map contains four cell-state channels plus one channel for each configured marker.

Cell Channels

Channel	Meaning
0	Unknown
1	Empty
2	Contact
3	Echo

Each cell belongs to exactly one of these four states.

Marker Channels

Each recognized marker has one dedicated channel.

Marker channels start at channel 4.

For marker index i:

channel = 4 + i

Channel	Meaning
4	Marker 0
5	Marker 1
6	Marker 2
...	...
4 + N - 1	Marker N - 1

5. State Flow

flowchart TD
    WM[World Model]

    WM --> RS[Robot Status]
    WM --> GM[Grid Map]
    WM --> MK[Markers]

    RS --> SENSOR[sensor]
    RS --> MOVE[canMoveStates]

    GM --> MAP[map]
    MK --> MAP

Loading

Summary

The Wheelly state provides:

• Orientation through sensor
• Movement and contact information through canMoveStates
• Spatial information through map

Together, these signals provide the Reinforcement Learning agent with information about the robot's orientation, movement constraints, and surrounding environment.