mentabotix

mentabotix

A dedicated lib to control 4-fixed-wheels robot

---

Installation

Use pdm to install mentabotix

# install pdm
python -m pip install pdm

# config pdm
pdm config pypi.url https://pypi.tuna.tsinghua.edu.cn/simple

# for stable release
pdm add mentabotix

# for unstable release
pdm add mentabotix --pre

Usage

Menta

You can use Menta to create the judge function closure of the robot, using the sensor data as the input.

Step 1: Define Sampler Functions

Firstly, you need to define sampler functions that adhere to the specifications outlined by the Menta class. These could include sequence samplers that return a series of data points, index samplers that provide data at a specific index, and direct response samplers that give an immediate value.

Sampler	Description	Type
mentabotix.SamplerType.SEQ_SAMPLER	The sensor data is returned in a Sequence.	Callable[[], Sequence[Union[float,int]]]
mentabotix.SamplerType.IDX_SAMPLER	Use a index to get the data.	Callable[[int], Union[float,int]]
mentabotix.SamplerType.DRC_SAMPLER	Direct read sensor data without any args.	Callable[[], Union[float,int]]

Do remember to add return type annotations to your sampler functions, which will be used to classify the samplers into their respective types in the Menta class.

def temperature_sequence_sampler() -> list[float]:
    """Simulates a sequence of temperature readings."""
    return [25.5, 26.0, 25.8]


def humidity_index_sampler(index: int) -> float:
    """Returns simulated humidity data at a given index."""
    humidity_values = [45.0, 50.0, 55.0]
    return humidity_values[index]


def light_direct_sampler() -> float:
    """Provides the current light intensity reading."""
    return 750.0

Step 2: Initialize Menta Instance and Add Samplers

After defining the sampler functions, initialize the Menta instance and add these samplers to it.

from mentabotix import Menta  # Ensure to import the correct Menta class


def temperature_sequence_sampler() -> list[float]:
    """Simulates a sequence of temperature readings."""
    return [25.5, 26.0, 25.8]


def humidity_index_sampler(index: int) -> float:
    """Returns simulated humidity data at a given index."""
    humidity_values = [45.0, 50.0, 55.0]
    return humidity_values[index]


def light_direct_sampler() -> float:
    """Provides the current light intensity reading."""
    return 750.0


menta_instance = Menta()
menta_instance.samplers.extend([temperature_sequence_sampler, humidity_index_sampler, light_direct_sampler])

Step 3: Update Sampler Types

Invoke the update_sampler_types method to automatically classify the samplers into their respective types.

from mentabotix import Menta  # Ensure to import the correct Menta class


def temperature_sequence_sampler() -> list[float]:
    """Simulates a sequence of temperature readings."""
    return [25.5, 26.0, 25.8]


def humidity_index_sampler(index: int) -> float:
    """Returns simulated humidity data at a given index."""
    humidity_values = [45.0, 50.0, 55.0]
    return humidity_values[index]


def light_direct_sampler() -> float:
    """Provides the current light intensity reading."""
    return 750.0


menta_instance = Menta()
menta_instance.samplers.extend([temperature_sequence_sampler, humidity_index_sampler, light_direct_sampler])

menta_instance.update_sampler_types()

Step 4: Construct Judge Function

To use the samplers in a meaningful way, you can construct an updater function that encapsulates a condition based on which system updates might occur. For example:

from mentabotix import SamplerUsage, Menta


def temperature_sequence_sampler() -> list[float]:
    """Simulates a sequence of temperature readings."""
    return [25.5, 26.0, 25.8]


def humidity_index_sampler(index: int) -> float:
    """Returns simulated humidity data at a given index."""
    humidity_values = [45.0, 50.0, 55.0]
    return humidity_values[index]


def light_direct_sampler() -> float:
    """Provides the current light intensity reading."""
    return 750.0


menta_instance = Menta()
menta_instance.samplers.extend([temperature_sequence_sampler, humidity_index_sampler, light_direct_sampler])

menta_instance.update_sampler_types()

# Define how samplers will be used
usages = [

    # Use sequence sampler for first and third temp data
    SamplerUsage(used_sampler_index=0, required_data_indexes=[0, 2]),
    # Use index sampler for the first humidity value
    SamplerUsage(used_sampler_index=1, required_data_indexes=[0]),
]

# Note: the judge source has some built-in syntaxes.Currently, sensor data indexes flattening and extra data context insertion
# s0 stands for the data at index 0 of the first sampler
# s1 stands for the data at index 2 of the first sampler
# s3 stands for the data at index 1 of the second sampler

judging_source = "(s0 > 25 and s1 < 50) or s3>baseline"

# Extra context for the judge function, here only contains the "baseline"
extra_context = {"baseline": 47}

from typing import Callable

# Construct the judge function object
updater_function: Callable[[], bool] = menta_instance.construct_inlined_function(usages, judging_source=judging_source,
                                                                                 extra_context=extra_context)

# Use the judge function to update the system
updater_function()


# Below is the equivalent implementation of the judge function, but is defined manually. 
# It acts exactly the same as the `updater_function` above.
def manual_judge_function() -> bool:
    """Manually construct the judge function."""
    seq_temp = temperature_sequence_sampler()
    return (seq_temp[0] > 25 and seq_temp[2] < 50) or humidity_index_sampler(0) > 47

In this case, a judge function closure is created using exec() method.Normally the built should have a better performance since the all the calls and variables are inlined and stored in the closure.

Botix

Welcome to the guide on using Botix for state-transition control schema creation and compilation. This document will walk you through the steps to design a control schema using state and transition concepts, and then how to compile those schemas into executable closures using the Botix framework.

Understanding State-Transition Control Schema

Let's start with the fundamental schema rules:

• A MovingState MUST connect exactly ONE MovingTransition as its input state.
• A MovingTransition MUST have at least ONE MovingState as either input or output.
• A State-Transition Control Schema MUST have EXACTLY ONE MovingState as its initial state and **AT LEAST ONE ** MovingState as its final state.
• A State-Transition Control Schema MUST NOT have any loop, a correct control schema should always be a tree graph.

Imagine you're designing an autonomous robot that needs to navigate different environments. Each behavior or action the robot can take is represented by a state, such as "moving forward," "turning left," or "halt." Transitions between these states are triggered by events or conditions, forming a control schema.

States

• In Botix, a state is represented by the MovingState class. Each state might have associated actions like setting motor speeds or changing direction.
• A state describe the robot's moving behavior, 4 motor speeds in this case

Transitions

• Transitions between states are handled by the MovingTransition class. They define how and when the robot moves from one state to another, possibly based on sensor inputs or internal conditions.
• Transitions describe how the robot moves from one state to another

Building Your Control Schema

Step 1: Define States

For each distinct behavior, create a MovingState instance. Include the actions or configurations that should occur in that state. For instance, creating a state for moving forward might look like this:

from mentabotix import MovingState

# the most basic syntax to control the robot advance in a straight line with speed of 10
# speed of all motors is 10
moving_forward = MovingState(10)
# Speeds for left and right motors are 10
moving_forward = MovingState(10, 10)
# Speeds for, in order, front_left, rear_left, rear_right, front_right motors are 10
moving_forward = MovingState(10, 10, 10, 10)

# Use assistant method to create the state instance
moving_forward = MovingState.straight(10)
moving_backward = MovingState.straight(-10)

turn_left = MovingState.turn("l", 10)
turn_right = MovingState.turn("r", 10)

diff_move_left = MovingState.differential("l", radius=20, outer_speed=70)
diff_move_right = MovingState.differential("r", radius=20, outer_speed=70)

drift_fl = MovingState.drift("fl", 50)
drift_rl = MovingState.drift("rl", 50)
drift_rr = MovingState.drift("rr", 50)
drift_fr = MovingState.drift("fr", 50)

stopped_state = MovingState.halt()

Step 2: Define Transitions

Transitions are created with MovingTransition. Specify the duration, any conditions under which the transition should occur (using a breaker function), and the source and destination states.

from mentabotix import MovingTransition, MovingState
from random import random

sensor_reading = lambda: random()  # Example sensor reading function
threshold = 0.6


def stop_condition() -> bool:  # the return type must be annotated, since an exception will be raised otherwise
    return sensor_reading() < threshold


moving_forward = MovingState(10)
stopped_state = MovingState(0)

transition_forward_to_stop = MovingTransition(
    duration=2,  # Duration to transition
    breaker=stop_condition,  # When to break the transition
    from_states=moving_forward,
    to_states=stopped_state,
)
# Such transition indicate the bot will move forward for 2 seconds OR stop when the sensor reading is below 25.0 

But some time you might want to create a transition that branches out to other states. For example, if the robot needs to move forward and then turn left. This can be done by creating a MovingTransition like this:

from mentabotix import MovingTransition, MovingState
from random import choice


def stop_condition() -> int:  # return type must be annotated
    return choice([0, 1, 2])


moving_left = MovingState.turn("l", 10)
moving_right = MovingState.turn("r", 10)
moving_dash = MovingState.straight(100)
stopped_state = MovingState(0)

transition_forward_to_stop = MovingTransition(
    duration=2,  # Duration to transition
    breaker=stop_condition,  # When to break the transition
    from_states=stopped_state,
    to_states={1: moving_left, 2: moving_right, 0: moving_dash},
)

# Such transition indicate the bot will stop for 2 seconds and then dash forward if nothing happens. ( means the top_condition never return a Non-Zero value)

# If the stop_condition do return a Non-Zero value, 
# the 2-seconds stop will be broken and the bot will either turn left or right according to the value returned by the stop_condition. (means if the value is 1, it will turn left, if the value is 2, it will turn right)

Step 3: Connect States with Transitions

Create a collection of your states and transitions, ensuring each transition correctly references its source and target states.

A legal control schema should have a start state and at least one end state. And each state can ONLY connect to ONE transition as its input state

from mentabotix import MovingState, MovingTransition
import random

moving_left = MovingState.turn("l", 10)
moving_right = MovingState.turn("r", 10)
moving_dash = MovingState.straight(100)
stopped_state = MovingState(0)

start_state = MovingState(100)


def _breaker_1() -> int:
    return random.choice([0, 1, 2])


def _breaker_2() -> int:
    return random.choice([0, 1])


transition_1 = MovingTransition(
    duration=2,  # Duration to transition
    breaker=_breaker_1,  # When to break the transition
    from_states=stopped_state,
    to_states={1: moving_left, 2: moving_right, 0: moving_dash},
)

transition_2 = MovingTransition(
    duration=2,  # Duration to transition
    breaker=_breaker_2,  # When to break the transition
    from_states=stopped_state,
    to_states={0: moving_dash, 1: stopped_state},
)

As the above example, transition_1 and transition_2 both have the stopped_state as the input state,which means the stopped_state connects more than ONE transition. Definition as such is illegal and will not pass the compile stage.

Normally, you can ensure that the structure is valid by using ensure_structure_validity. to eliminate this problem.

from mentabotix import Botix

# A StructuralError will be raised if the structure is not valid
Botix.ensure_structure_validity([transition_1, transition_2])

Compiling to Closures with Botix

Once your control schema is defined, Botix can help you compile this structure into executable code, often referred to as closures, which can directly control the robot's behavior.

Step 4: Compile the Schema

Botix provides methods to compile your states and transitions into a Match-Case-Structured Closure.

from mentabotix import MovingState, MovingTransition, Botix
from bdmc import CloseLoopController, MotorInfo

con: CloseLoopController = CloseLoopController(motor_infos=[MotorInfo(code_sign=3, direction=1),
                                                            MotorInfo(code_sign=1, direction=1),
                                                            MotorInfo(code_sign=2, direction=-1),
                                                            MotorInfo(code_sign=4, direction=-1)], port="COM3")

start_state = MovingState(-1)
state_a = MovingState(0)
state_b = MovingState(1)
state_c = MovingState(2)
# 创建一些假的MovingTransition对象用于测试
transition_start_a = MovingTransition(duration=0.1, from_states=start_state, to_states=state_a)
transition_ab = MovingTransition(duration=1, from_states=state_a, to_states=state_b)
transition_bc = MovingTransition(duration=2, from_states=state_b, to_states=state_c)

botix = Botix(controller=con)
(botix
 .append_token(transition_start_a)
 .append_token(transition_ab)
 .append_token(transition_bc))

# not compile the code into closure, just return the code lines and context, which is human readable.
compiled_code_lines, variables_context = botix.compile(return_median=True)

print(compiled_code_lines)

By printing out the compiled_code_lines list, you can see the code lines that are generated by Botix.

[
    "def _func():",
    "    con.set_motors_speed((-1, -1, -1, -1)).delay(0.1).set_motors_speed((0, 0, 0, 0)).delay(1).set_motors_speed((1, 1, 1, 1)).delay(2).set_motors_speed((2, 2, 2, 2))",
]

But usually, you want to compile the code into a closure that can be used to control the robot with higher performance and less human error.

from mentabotix import MovingState, MovingTransition, Botix
from bdmc import CloseLoopController, MotorInfo
from typing import Callable

con: CloseLoopController = CloseLoopController(motor_infos=[MotorInfo(code_sign=3, direction=1),
                                                            MotorInfo(code_sign=1, direction=1),
                                                            MotorInfo(code_sign=2, direction=-1),
                                                            MotorInfo(code_sign=4, direction=-1)], port="COM3")

start_state = MovingState(-1)
state_a = MovingState(0)
state_b = MovingState(1)
state_c = MovingState(2)
# create some transitions
transition_start_a = MovingTransition(duration=0.1, from_states=start_state, to_states=state_a)
transition_ab = MovingTransition(duration=1, from_states=state_a, to_states=state_b)
transition_bc = MovingTransition(duration=2, from_states=state_b, to_states=state_c)

botix = Botix(controller=con)
(botix
 .append_token(transition_start_a)
 .append_token(transition_ab)
 .append_token(transition_bc))

# not compile the code into closure, just return the code lines and context, which is human-readable.
function_closure: Callable[[], None] = botix.compile(return_median=False)

print(function_closure)

By printing out the function_closure object, you can see the compiled code as a closure that can be called

< function _func at 0x0000020D40EAECA0 >

Usage is as follows

# call the closure, which will execute the compiled code
function_closure()

Of course, you can also build a closure with branching logic in it.

from mentabotix import MovingState, MovingTransition, Botix
from bdmc import CloseLoopController, MotorInfo
import random
from typing import List, Tuple, Dict, Any, Callable

controller: CloseLoopController = CloseLoopController(
    motor_infos=[MotorInfo(code_sign=3, direction=1),
                 MotorInfo(code_sign=1, direction=1),
                 MotorInfo(code_sign=2, direction=-1),
                 MotorInfo(code_sign=4, direction=-1)], port="COM3"
)

botix = Botix(controller=controller)
state_a = MovingState(100)
state_b = MovingState(200)
state_c = MovingState(300)
state_d = MovingState(400)
state_e = MovingState(500)
state_f = MovingState(600)


def transition_breaker_fac(lst: List[int]):  # a simple tool function to create the breaker
    def _inner() -> int:
        return random.choice(lst)

    return _inner


transition_a_bcd = MovingTransition(
    duration=1,
    from_states=state_a,
    to_states={0: state_b, 1: state_c, 2: state_d},
    breaker=transition_breaker_fac([0, 1, 2]),
)
transition_d_ef = MovingTransition(
    duration=1,
    from_states=state_d,
    to_states={2: state_e, 1: state_f},
    breaker=transition_breaker_fac([0, 1]),
)

botix.extend_pool([transition_a_bcd, transition_d_ef])

compiled: Tuple[List[str], Dict[str, Any]] = botix.compile(
    return_median=True)  # not compile the code into closure, just return the code lines and context, which is human readable.
assert (
    [
        "def _func():",
        "    match con.set_motors_speed((100, 100, 100, 100)).delay_b_match(1,transition0_breaker_1,0.01):",
        "        case 0:",
        "            con.set_motors_speed((200, 200, 200, 200))",
        "        case 1:",
        "            con.set_motors_speed((300, 300, 300, 300))",
        "        case 2:",
        "            match con.set_motors_speed((400, 400, 400, 400)).delay_b_match(1,transition1_breaker_1,0.01):",
        "                case 0:",
        "                    con.set_motors_speed((500, 500, 500, 500))",
        "                case 1:",
        "                    con.set_motors_speed((600, 600, 600, 600))",
    ] == compiled[0],
)

# compile to closure
compiled_closure: Callable[[], None] = botix.compile(return_median=False)

# call the closure, which will execute the compiled code
compiled_closure()

Schema Visualization

Botix support schema visualization with export_structure method, which write Puml source code to a file.

from mentabotix import MovingState, MovingTransition, Botix
from bdmc import CloseLoopController, MotorInfo
from typing import List
import random

# init the state-transition schema
state_a = MovingState(100)
state_b = MovingState(200)
state_c = MovingState(300)
state_d = MovingState(400)
state_e = MovingState(500)
state_f = MovingState(600)


def transition_breaker_fac(lst: List[int]):
    def _inner() -> int:
        return random.choice(lst)

    return _inner


transition_a_bcd = MovingTransition(
    duration=1,
    from_states=state_a,
    to_states={1: state_b, 2: state_c, 3: state_d},
    breaker=transition_breaker_fac([1, 2, 3]),
)
transition_d_ef = MovingTransition(
    duration=1,
    from_states=state_d,
    to_states={1: state_e, 2: state_f},
    breaker=transition_breaker_fac([1, 2]),
)

# make the botix object
botix = Botix(controller=CloseLoopController(motor_infos=[MotorInfo(i) for i in range(4)], port="COM3"))

# add the transition
botix.token_pool.extend([transition_a_bcd, transition_d_ef])

# export the structure
Botix.export_structure("schema.puml", botix.token_pool)

The result will be written to schema.puml and below is the expected Puml source code.

@startuml
state "5-MovingState(600, 600, 600, 600)" as state_6
state "4-MovingState(500, 500, 500, 500)" as state_5
state break_2 <<choice>>
note right of break_2: _inner() -> int
state "3-MovingState(400, 400, 400, 400)" as state_4
state "2-MovingState(300, 300, 300, 300)" as state_3
state "1-MovingState(200, 200, 200, 200)" as state_2
state break_1 <<choice>>
note right of break_1: _inner() -> int
state "0-MovingState(100, 100, 100, 100)" as state_1
state_1 --> break_1
break_1 --> state_2: 1
break_1 --> state_3: 2
break_1 --> state_4: 3
state_4 --> break_2
break_2 --> state_5: 1
break_2 --> state_6: 2

[*] --> state_1

state_2 --> [*]
state_3 --> [*]
state_5 --> [*]
state_6 --> [*]

@enduml

Below is the expected render result:

Use State-Transition Composer

State-Transition Composer is a tool that helps you to design and compile state-transition control schema.

from mentabotix import MovingChainComposer, MovingState, MovingTransition, Botix
from bdmc import CloseLoopController, MotorInfo

# init the state-transition composer
comp = MovingChainComposer()

# add some states and transitions one by one to the composer, the composer will auto-connect the states and transitions
(comp
 .add(MovingState(0))
 .add(MovingTransition(0.2))
 .add(MovingState(1000))
 .add(MovingTransition(0.3))
 .add(MovingState(2000)))

# export the structure
states, transitions = comp.export_structure()

# let's use botix to make the visualization!
# first make the botix object
con = CloseLoopController(motor_infos=[MotorInfo(i) for i in range(4)])
botix = Botix(controller=con)

# make the visualization
Botix.export_structure("composed.puml", transitions=transitions)

The exported structure will be written to composed.puml, and below is the expected Puml source code.

@startuml
state "3-MovingState(2000, 2000, 2000, 2000)" as state_3
state "2-MovingState(1000, 1000, 1000, 1000)" as state_2
state "1-MovingState(0, 0, 0, 0)" as state_1
state_1 --> state_2
state_2 --> state_3

[*] --> state_1

state_3 --> [*]

@enduml

The render result is shown below:

Logging

use set_log_level to silent the console to improve the performance in high pressure conditions

from mentabotix import set_log_level

"""
Logging DEBUG - Debugging information, used for detailed development phase logs, typically with a value of 10.
Logging INFO - Information message, used to inform the general program running status, with a value of 20.
Logging WARN - A warning message indicating that there may be a problem but the program is still running, with a value of 30.
Logging Error - Error message indicating an issue preventing the program from executing properly, with a value of 40.
Logging CRITICAL - Fatal error message indicating a serious system failure with a value of 50.
"""

set_log_level(50)  # set the log-level to 50, which makes logger only print the msg important than the CRITICAL logging

from logging import CRITICAL

set_log_level(CRITICAL)  # this has the same effect as above